CEU Electronic Theses and Dissertations, 2016
| Author | Horváth, Imola |
|---|---|
| Title | Empirical Forecasting Model of Hungarian Power System Load (2014-2015) |
| Summary | Electricity cannot be stored such as the other commodities, therefore the demand and the supply side of the power market has to be in balance all the time. As a result, an accurate forecasting model is crucial in terms of system security in order to avoid blackouts or excessive power generation. However, the currently applied model of Hungarian Transmission System Operator is inaccurate, which resulted in a high operational cost. The main objective of my thesis is to develop a handy and accurate system load forecasting model for 2015. For my research, I used the database of the Hungarian Transmission System Operator, which contains hourly system load and weather forecast data for 2014-2015. I tested the forecasting performance of double seasonal ARIMA, Holt-Winters exponential smoothing with double seasonal cycle and log-linear models. One of the key findings of my thesis is that in terms of the Hungarian system load the dummy variables can handle the seasonal pattern better than the seasonal differencing. Moreover, the calendar variables proved to be more significant explanatory variable, than the weather variables, especially those which were responsible for the effect of the holidays. Finally the log-linear model specification had the best forecasting accuracy, in the case of which the average MAPE of a day-ahead forecast is 2.4%. With the help of my model the Hungarian Transmission System Operator could increase the predictability and security of the system, furthermore decrease its operational cost. |
| Supervisor | Gábor Kőrösi |
| Department | Economics MA |
| Full text | https://www.etd.ceu.edu/2016/horvath_imola.pdf |
Visit the CEU Library.
© 2007-2021, Central European University