Correction of the parametric model of the Day-Ahead Market system using the Artificial Neural Network

Abstract

The paper shows that it is possible to correct the identification model of the Day-Ahead Market system by employing the Perceptron Artificial Neural Network. First, a simulation model of the DAM system at the POLPX has been built, and then it has been shown how the model can be corrected so that the weighted average electricity prices obtained are close enough to the exchange-quoted ones. Next, simulation, comparative and sensitivity studies of the model were carried out for forecast data for four characteristic hours: 6, 12, 18, and 24 of the following year. Many interesting research results were obtained, including a result of sensitivity testing it was shown that the obtained models can be used in forecasting studies.