以ARMA及GM(1,1)模型預測風力發電量之研究

Using an ARMA & GM(1,1) model to predict electrical production of wind power

趙自偉(Tzu-Wei Chao)

指導教授 ： 王順成 ； 白子易

朝陽科技大學/理工學院/環境工程與管理系/碩士(2012年)

摘要

摘要
本研究應用ARMA(Autoregression-Moving Average)模型及GM (Grey Model) (1,1)模型預測「四湖風力發電場及麥寮風力發電場」之發電量。通常運用AR(Autoregression)或MA(Moving Average)配適模型時，需要較多參數才能配適出合適的模型，為了克服此一困難點，利用ARMA模型或許可以精簡配適模型之參數；而灰色系統理論則是利用系統少數 (至少4個)之輸出資料建立灰色模型 (GM) 近似灰色系統之動態行為，進而對此少量輸出數資料之系統進行灰預測 (Grey Prediction) 。研究結果顯示，四湖風力發電場ARMA模型預測時的MAPE(Mean Absolute Percentage Error)介於68.6217至93.7750之間；預測時的R值介於-0.16至-0.08之間。麥寮風力發電場ARMA模型預測時的MAPE介於140647.3501至167633.3169之間；預測時的R值介於0.67至0.69之間。四湖風力發電場GM (1, 1) 模型預測時的MAPE介於37.9580至65.2057之間；預測時的R值介於0.14至0.89之間。麥寮風力發電場GM (1, 1) 模型預測時的MAPE介於154.4582至193.7002之間；預測時的R值介於0.26至0.79之間。

關鍵字

發電量 ； 麥寮 ； GM(1，1)模型 ； 四湖 ； ARMA模型

並列摘要

Abstract
This study used ARMA (Autoregression-Moving Average)model and GM (Grey Model) (1,1) model to predict the power output of Si-Hu Wind Power Plant and Mai-Liao Wind Power plant. Generally, when using AR(Autoregression) or MA(Moving Average) to fit models, multiple parameters are required to fit an appropriate model; thus, in order to overcome this problem, the ARMA model may simplify the parameters of fit model. The grey system theory uses a few (at least 4) output data of system to build Grey Model (GM) approximating the dynamic behavior of grey system, so as to carry out Grey Prediction for the system of a few output data. The results showed that the MAPE(Mean Absolute Percentage Error) of Si-Hu Wind Power Plant in the ARMA model prediction is 68.6217 to 93.7750; and the R value in prediction is -0.16 to -0.08. The MAPE of Mai-Liao Wind Power Plant in ARMA model prediction is 140647.3501 to 167633.3169; and the R value in prediction is 0.67 to 0.69. The MAPE of Si-Hu Wind Power Plant in GM (1, 1) model prediction is 37.9580 to 65.2057; and the R value in prediction is 0.14 to 0.89. The MAPE of Mai-Liao Wind Power Plant in GM (1, 1) model prediction is 154.4582 to 193.7002; and the R value in prediction is 0.26 to 0.79.

並列關鍵字

Mailiao ； Electrical Production ； GM(1，1) model ； ARMA model ； Sihu

參考文獻

吳繼平，「應用類神經網路及基因演算法預測風速與風力發電量」，碩士論文，中原大學電機工程研究所，中壢(2007)。

郭瑞玲，2011，以7種GM (1, 1) 模型預測舊濁水溪水質之研究，朝陽科技

廖大榮，2012，以ARMA模型預測太陽能電廠發電量之研究，朝陽科技大

Chang S.C., Pai T.Y., Ho H.H., Leu H.G. and Shieh Y.R. Evaluating Taiwan’s air quality variation trends using grey system theory. Journal of the Chinese Institute of Engineers, 30(2), 361-367 (2007).

Pai T.Y., Ouyang C.F., Su J.L. and Leu H.G. Modelling the steady-state effluent characteristics of the TNCU process with ASM2d under varied SRT conditions. Journal of the Chinese Institute of Environmental Engineering, 10(1), 35-42 (2000a).