부산대학교 도서관

This paper presents a forecasting approach for national annual peak electricity demand. Forecasting is performed by rescaling observed historic demand. The rescaled peaks are then used to produce probabilistic forecasts of peak demand for a target year. Both the rescaling approach and the probabilistic forecasting methodology are carefully validated and shown to provide a good fit to the data. Comparisons are made with the National Grid’s Average Cold Spell (ACS) Methodology which aims to estimate the level of electricity demand in Great Britain for which there is a 50 percent probability of exceedance in a given year.