Day-ahead deregulated electricity market price forecasting using neural network input featured by DCT

• We presented DCT input featured FFNN model for forecasting in Spain market.
• The key factors impacting electricity price forecasting are historical prices.
• Past 42 days were trained and the next 7 days were forecasted.
• The proposed approach has a simple and better NN structure.
• The DCT-FFNN mode is effective and less computation time than the recent models.

In a deregulated market, a number of factors determined the outcome of electricity price and displays a perplexed and maverick fluctuation. Both power producers and consumers needs single compact and robust price forecasting tool in order to maximize their profits and utilities. In order to achieve the helter–skelter kind of electricity price, one dimensional discrete cosine transforms (DCT) input featured feed-forward neural network (FFNN) is modeled (DCT-FFNN). The proposed FFNN is a single compact and robust architecture (without hybridizing the various hard and soft computing models). It has been predicted that the DCT-FFNN model is close to the state of the art can be achieved with less computation time. The proposed DCT-FFNN approach is compared with 17 other recent approaches to estimate the market clearing prices of mainland Spain. Finally, the accuracy of the price forecasting is also applied to the electricity market of New York in year 2010 that shows the effectiveness of the proposed DCT-FFNN approach.