Short-term electricity price forecasting using artificial neural networks: a case study of Russian wholesale electricity market.
The electricity market liberalization in Russia has led to the emergence of the wholesale power market. Since that time, market participants operate in the competitive environment, facing everyday with market strategy planning issues. Under these conditions forecasting electricity prices has become an integral and daily challenge for most market participants. This is also true in case of high uncertainty that typically characterizes Russian electricity market. To this end, it is especially demanded to formulate and apply precise forecast models to predict market conjecture.
In this paper we consider the possibility of using neural networks for short-term electricity price forecasting on the Russian day-ahead market based only on market specific deterministic factors. The results show that the proposed set of six factors accurately describe the market conjecture and proposed model allows to get reliable month hourly price forecast in four different seasons of the year. The proposed model shows the lowest average prediction error rates for each hour of the month and, in turn, allows market participant to anticipate significant deviations of the price.