단기 천연가스 가격 예측에 있어서 인공신경망의 활용에 대한 연구

초록

Because of its good performance in forecasting, artificial neural network(ANN) has been steadily appliedin time series analyses. We applied this ANN to forecast short-run natural gas prices which received a lot ofattention due to the boom of unconventional gas developments. The Henry Hub spot price was analyzed anda back-propagation algorithm which is the most popular method in ANN was applied. We changed ANNforecasting methods(cascading vs. fixed point), training periods, input numbers, and nodes of hidden layer fora comparison. As result, the cascading model with six inputs and ten nodes in the hidden layer had the minimumforecasting error which means the optimal model.

키워드