Neural and Fuzzy Neural Networks in Prediction of Natural Gas Consumption

In this work several approaches to prediction of natural gas consumption with neural and fuzzy neural systems are analyzed and tested. The data covers daily natural gas load in two different regions of Poland. Prediction strategies tested in the paper include: single neural net module approach, combination of three neural modules, temperature context based method, and application of fuzzy neural networks. The results indicate the superiority of temperature context based method and the modular approach over single neural net and fuzzy neural approaches. One of the interesting issues observed in the paper is relatively good performance of tested methods in the case of long-term (four week) prediction compared to mid-term (one week) prediction. Generally, the results are superior to those obtained by linear and quadratic regression models and by statistical methods currently used for this task in the gas company under consideration.