Electricity price forecasting using neural networks and similar days

Forecasting electricity prices is one of the most important issues in the competitive environment of the power industry. The price of electricity influences important decisions by market participants and plays a crucial role in establishing a proper economical operation. The main objective of an electricity market is to reduce the cost of electricity through competition [1–5]. Nowadays, electricity has been turned into a traded commodity, to be sold and bought at market prices. In general, loads and prices in wholesale markets are mutually intertwined activities [3]. However, electricity has its own distinct characteristics since it cannot be queued and stored economically with the exception of pumped-storage hydro plants when appropriate conditions are met, and the power system stability requires a constant balance between generation and load [1, 6, 7]. Furthermore, electricity is affected by power network limitations. In many cases, electric power cannot be transported from one region to another because of existing bottlenecks or limited transmission capacity of the grid. Hence, prices are local and differ among regions [2, 6, 8]. Electricity price can rise to tens or even hundreds of times of its normal value showing one of the greatest volatilitiesamong all commodities. Application of forecasting methods commonly used in othercommodity markets can have a large error in forecasting the price of electricity