A Combined Seasonal ARIMA and ANN Model for Improved Results in Electricity Spot Price Forecasting: Case Study in Turkey


OZOZEN A., KAYAKUTLU G., KETTERER M., Kayalica O.

Portland International Conference on Management of Engineering and Technology (PICMET), Hawaii, United States Of America, 4 - 08 September 2016, pp.2681-2690 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • City: Hawaii
  • Country: United States Of America
  • Page Numbers: pp.2681-2690

Abstract

Developing countries arc trying to improve the competitiveness of the energy markets with continuous liberalization. This makes the market highly sensitive. Every player in the market has a greater need to know about the smallest change in the market. Hence, ability to see what is ahead is a valuable advantage to make the right move. A time series forecasting with the smallest errors would he a powerful tool for the energy producers. This paper proposes combined methodology in time series forecasting. Generally accepted and widely used ARIMA and ANN with backpropagation learning are combined. The methodology is implemented for the day ahead Turkish power market. It is observed that the proposed methodology gives results with reduced errors. The achievements are compared with conventional use of both ARIMA and ANN.