- 學位論文
解制電業市場區域邊際電價之預測
Locational Marginal Price Forecasting in Deregulated Power Markets
摘要
隨著電力市場的解制而自由化後，區域邊際價格（Locational Marginal Prices, LMPs）預測變的重要了。而更精準的預測出未來之電價，可提供電力市場參與者做出正確的經濟判斷，來減少投資風險，亦可增進市場的效率，並降低成本及能源價格。本論文利用類神經網路與模糊推論提出一個預測LMP的方法。 本論文利用模糊規則的方法，在系統發生故障時，考慮時間、負載和LMP差（LMP 值由前一小時與現在時間的變化量），並且進行預測未來LMP節點價錢，而且配合遞迴式類神經網路，將完整資料分成工作日與週末進行LMP預測。 本論文將運用美國PJM（Pennsylvania，New Jersey，Maryland）Interconnection系統的歷史資料作為模擬測試之對象，利用所提出的人工智慧方法進行LMP之預測，並計算實際值與預測值兩者之誤差量及相關係數，以證實本論文所提出之LMP預測方法之可行性。
並列摘要
The Locational Marginal Price (LMP) forecasting becomes important due to the deregulation in the electricity market. Owing to the proper forecasting in price, the market participants may make accurate economic decision to diminish the investment risk, reduce the cost and or energy price. Therefore, thesis presents a method using the artificial neural network (ANN) and fuzzy reasoning for forecasting the LMP. Taking the system contingencies into account, this thesis employs the fuzzy reasoning considering the factors of time, load, and LMP variation for obtaining the information of LMP trend. This will serve as an input for the recurrent ANN for forecasting the weekday and weekend LMPs. The PJM realistic historical data were used for verifying the proposed method. The correlation and errors between the forecasting and realistic data were given for showing the applicability of the proposed method.