以人工智慧進行解制市場區域邊際電價之預測

Locational Marginal Price Forecasting in Deregulated Markets Using Artificial Intelligence

蕭全佑(Chuan-Yo Hsiao)

指導教授 ： 洪穎怡

中原大學/工學院/電機工程研究所/碩士(2001年)

https://doi.org/10.6840/cycu200100014

摘要

摘 要

近來解制及自由化在世界各地展開，不論在電信、運輸或電力產業均已見其成效，而電力產業邁向自由化後，市場買賣的行為變成一種透過競標方式的商業交易行為。在現今解制電業市場中，區域邊際價格（Locational Marginal Prices, LMPs）預測變得越顯重要，推估LMPs短期走勢，可幫助市場參與者於現貨價格市場中制訂競標策略。準確地預測電能價格是相當關鍵的，因為預測準確性的增加，相對之下可以降低發電業者或配電公司因為高估或低估自己的收益所造成的風險，因此也可提供較佳的風險管理。本論文利用類神經網路與模糊分類技術提出一個預測LMP的方法。

本論文利用模糊分類（Fuzzy-C-Means, FCM）的方法，對負載時段分別進行輕載、重載兩類，以及輕載、中載、重載三類的分割，並且配合遞迴式類神經網路與傳統倒傳遞類神經網路，分別對完整資料以及工作日與假日分開進行預測。

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。

關鍵字

節點電價預測 ； 模糊分類 ； 類神經網路

並列摘要

Abstract

Recently, deregulation has had a great impact on the telecommunications, transportation and electric power industry in various countries. Bidding competition is one of the main transaction approaches after deregulation. In today’s deregulated markets, Locational Marginal Prices (LMP) forecasting is becoming more and more important. In the short-term, LMPs reveal important information that is helpful for the market participants to develop their bidding strategies. Accuracy in forecasting these LMPs is crucial, since more accuracy in forecasting reduces the risk of under- or over-estimating the revenue from the gencos or transcos and provides better risk management. Artificial Neural Network (ANN) and Fuzzy-C-Means (FCM) algorithm based LMP forecasting is presented in this thesis..
The data are partitioned into two clusters (low load and heavy load) and three clusters (low load, middle load and heavy load) by FCM according to the load levels. The performance on LMPs forecasting toward the whole data and the data with separating weekday data from weekend data by Recurrent Neural Network (RNN) and Back-Propagation Network is given.
This thesis employs the PJM (Pennsylvania, New Jersey, Maryland) Interconnection historical data to serve as a test system for LMPs forecasting by the proposed Artificial Intelligence method. The errors and correlation coefficient between actual LMPs and forecasted LMPs are shown in order to show the applicability of the proposed method.

Keywords: Artificial Neural Network, Fuzzy-C-Means, LMP forecasting

並列關鍵字

artificial neural network ； Ruzzy-C-Means ； LMP forecasting

參考文獻

[1] Y.Y. Hong, and M.T. Weng, “Investigation of Nodal Prices in Deregulated Competitive Market -Case Studies,” international conference on PowerTech, Budapest, Aug. 29-Sep. 2, 1999, pp. 161.

[4] S. Hunt and G. Shuttleworth, Competition and Choice in Electricity, John Wiley & Sons, New York, 1996.

[5] T.W. Hissey, "International Experience in Transmission Open Access," IEEE Power Engineering Review, Vol. 14, No. 12, Dec. 1994, pp.3-18.

[7] J.W. Williams, "Open Transmission Access," IEEE Power Engineering Review, Vol. 16, No. 5, May 1996, pp.17-19.

[10] "Capacity Reservation Open Access Transmission Tariffs," Federal Energy Regulatory Commission, Docket No. RM 96-11-000, 1996.