Studies on Several Issues of Electricity Price Forecast

Abstract

The main models used in electricity price forecast have been reviewed in this paper. Several statistical models based on machine learning algorithms have been tested and compared to forecast the next-day hourly electricity price in Nordpool electricity market. Stationarity of electricity price has also been studied and it is concluded that electricity price is daily seasonal. Many features of both local region and neighboring regions from the past week have been respectively analyzed in terms of the importance to electricity price forecast. Results show that local zone, compared with neighbor zones, exerts the major influence on electricity price. Meanwhile, price is the most important category other than categories like demand, wind generation or transmission congestion rate. Hydro reserve is also an important parameter to electricity price forecast.