The authors illustrate an application of a recently developed stochastic modeling and analysis methodology, called data dependent systems (DDS), to short-term load forecasting. The salient features of DDS methodology pertaining to forecasting and updating the forecasts are briefly reviewed. Univariate as well as simplified vector models called extended autoregressive moving average (EARMA) models are obtained for actual hourly load and weather data for a small community (Curtis, Nebraska). The conditional expectation of the statistically adequate ARMA and EARMA models provides an accurate forecast for peak values of the load. The usefulness of weather data as leading indicators is explored. The improvements due to updating and using leading indicators are also discussed.
|Original language||English (US)|
|Number of pages||5|
|Publication status||Published - Dec 1 1985|
ASJC Scopus subject areas