http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
신동환(Donghwan Shin),백승엽(Seungyup Baek),이율호(Yulho Lee),강상규(Sanggyu Kang) 대한기계학회 2018 대한기계학회 춘추학술대회 Vol.2018 No.12
A latest korea electricity market participants are compensated their profits by system marginal price (SMP) determined at the korea power exchange (KPX). SMP is the most expensive price of an operated power generator at each hour and which is influenced by many external factor. SMP forecasting is important for profit maximization of power generation businessman. In this study, a SMP time series data forecasting model have been developed using MATLAB Neural Net Time Series<SUP>®</SUP> to predict the short-term SMP. preprocessing has been conducted to decrease a mean squared error (MSE) by comparing with gradient of present value and previous value. Moving average is used as an exogenous variable and nonlinear autoregressive exogenous (NARX) network is used for data training. One month forecasting along the moving average change has been compared. Training model shows that the MSE is influenced strongly by the preprocessing and it can be decreased by iteration in the filtering process. For more accuracy forecasting, another exogenous variables have to be investigated such as global oil price, tax, electricity demand. Another filtering and training methods also have to be compared for model optimization.