Comparing Neural Networks in Forecasting Market Electricity Prices and Regional Energy Consumption

The paper compares popular architectures of artificial neural networks used in forecasting market electricity prices and the total regional energy consumption. Various structures of neural networks used in forecasting have been considered. These are multilayer perceptron and popular architectures of deep neural networks, as well as their combinations. The structures of input datasets for forecasting using chosen neural networks have been described. To compare various neural networks, the authors performed many numerical experiments, the results of which are presented in the paper. Based on these results, researchers may choose a neural network structure, which allows improving the reliability and accuracy of forecasting market electricity prices and the total regional energy consumption.