Forecasting Model of Electricity Demand in the Nordic Countries

Sammanfattning: A model implemented in order to describe the electricity demand on hourly basis for the Nordic countries. The objective of this project is to use the demand data simulated from the model as input data in the price forecast model, EMPS model, at Vattenfall. The time horizon is 5 years, 6 years including the current year. After different models tried out, the final model is described by fundamental and autoregressive time variant variables, an ARX model. The variable of temperature is described by historical data from 46 years which are used to create an idea of the outcome variation depending on the weather. Non parametric bootstrap of the residuals is used when adding noise to the simulation. The ARX parameters was estimated by prediction error method but a two-step estimation was also tried, by first estimating the fundamental parameters and then model the rest of the demand by an AR process. The second method was supposed to increase the weight on the fundamental variables. Results of the simulation Indicates of a realistic description of the electricity demand which is an improvement of the earlier demand input to the EMPS model but the difference is not always seen in the outcome of the price forecast. The results are discussed in Chapter 5.