engleski

Predicting natural gas consumption by neural networks

The aim of the paper is to create a prediction model of natural consumtpion on a regioanal level by using neural networks, and to analyze the results in order to improve prediction accuracy in further research. The outputvariable consisted of the next-day gas consuption in hourly intervals, while the input space included previous-day consumption in addition to exogenus variables. After conducting a feature selection procedure, two neural network algorithams were trained and tested: the multilayer perceptron and the radial basis function network with different activation functions. The dataset consisted of real historical data of Croatian gas distributor. The best neural network model is selected on the basis of the mean absolute percentage error obtained on the test sample. The results were analyzed, and some critical hours and days were identified. Gudidelines were reported that could be valuable to both researchers and practitioners in this area.

natural gas consumption; neural networks; multilayer perceptron; radial basic function network; fuzzy variable

nije evidentirano