Forecasting Power Load Demand Using Holt-Winters Model

Alfeo  B. Tulang,; Alwielland  Q. Bello,

Forecasting Power Load Demand Using Holt-Winters Model

Abstract:

Virtually time series observations such as power load demand follow some seasonal patterns. Electric service utilities depend on most quantitative forecasting models. This paper presented forecasting procedure using the Holt-Winters exponential smoothing model. Since data were stochastic and assumed to follow a multiplicative seasonality, this study adopted the classical decomposition method in the analysis of seasonal data. Decomposition is an approach of separating the time series into its component parts. The components of the Holt-Winters model included the level, trend, the seasonal index, and randomness with some smoothing constants which were estimated. Power load demand data were categorized as commercial, residential, industrial or others, from a given area of a power utility. The purpose of this study was to develop long-term forecasts of the power load demand in Cagayan de Oro City as well as to promote the use of Holt-Winters model as a time series forecasting method. With some numerical results, the Holt-Winters model predicted the patterns of power demand over a given time interval.