Load forecasting in a power system from a supply authority point of view

✍ Scribed by A.C. Tsoi; M.U. Kobe

• Publisher: Elsevier Science
• Year: 1983
• Tongue: English
• Weight: 996 KB
• Volume: 6
• Category: Article
• ISSN: 0378-7796
• DOI: 10.1016/0378-7796(83)90016-0

✦ Synopsis

The ability to predict accurately the shortterm demand on a power system is important to its optimal operation. We have studied halfhourly controlled demand data, covering a six-month period, with the aim of predicting morning and evening peak demands, both of which do not always occur at a fixed time. The Box-Jenkins method applied to the deperiodised and deseasonalised data was used to forecast half an hour ahead. The resulting prediction accuracies (to within +5%)depended critically on the amount of data used in determining the model. The temperature variations were correlated with the controlled demand data, and an improved forecast was obtained. Also, the base load was approximately separated from the given data, and a forecast was obtained based on the prediction of the temperature-induced load. The problem of restoring the uncontrolled demand curve has been studied, and a forecast was obtained for the uncontrolled demand peaks. The applications of such forecasts in relation to load-shedding strategies is also discussed.