應用適應性類神經模糊推論系統於冰水主機耗電量之預測

Application of Adaptive Neuro-Fuzzy Inference System to Predict Chiller Power Consumption

林良興(Liang-hsing Lin)

指導教授 ： 張永宗

國立臺北科技大學/機電學院/能源與冷凍空調工程系碩士班/碩士(2008年)

摘要

現今科技工業相互競爭之世代，大量的電力消耗是必然之趨勢，該如何節省能源是相當重要的課題。為了得知冰水主機電力需求負荷，傳統方式以線性迴歸建立主機耗電模型。而本研究應用適應性類神經模糊推論系統建立主機耗電模型，提高冰水主機的耗電量預測之精準度，此方法解決了傳統模糊系統上，以專家的經驗定義模糊規則庫的困難度；另一方面，也解決類神經網路系統在初始時需決定的隱藏層神經元數目。
利用適應性類神經模糊推論系統，將各台冰水主機運轉數據中的冰水出入水溫及冷卻水出入水溫等四個影響參數同時考慮時，所得到的預測結果相當準確，與實際耗電量相比，各台冰水主機耗電量模型準確率高達99％以上，再將相同冰水進出水溫度、冷卻水進出水溫度，代入各台冰水主機模型，我們可得到在相同情況下各台冰水主機的耗電量，取三台耗電量最低的冰水主機當作最佳冰水主機排序（Optimal Chiller Sequencing，簡稱OCS）所使用的冰水主機。透過此方法可發現耗電量有顯著的減少，在綠色節能方面提供了一種節能方式。

關鍵字

適應性類神經模糊推論系統 ； 冰水主機 ； 模糊理論

並列摘要

It’s a necessary to use considerable electricity in order to compete in science and technology among each industry, and how to save energy become very important in nowadays. Usually we apply linear regression equation to model power consumption of chiller for the purpose of to know the power consumption of chiller. In this research, I apply adaptive neuro-fuzzy inference system to predict chiller power consumption, this method can improve the accuracy of predicting power consumption of chiller, and this method not to adopt the master's experience fuzzy rule base of traditional fuzzy system; and can’t try and error to find the quantity of hidden layer neuron as same as neuron network system.
We have to consider the practical operation date of each chillers that is the input and output cooling water temperature and input and output chilling water temperature, using those date into the adaptive neuro-fuzzy inference system, we can find the predicting result is very accuracy. The predicting power consumption of chiller contrast with the practical power consumption of chiller, we can find the analogy between the predicting and the practical get up to 99%. And then we put practical operation date into the power consumption modeling of chiller, to get the predicting power consumption of chiller. Finally, to contrast with the power consumption of each chiller and there are three chillers that we choose have less power consumption than others. The three chillers that we choice can work with other chiller to accomplish optimal chiller sequencing. This method is effective in reducing the power consumption.

並列關鍵字

ANFIS ； Chiller ； Fuzzy Theory

參考文獻

[14] 戴邦文，應用於節能績效保證合約之冰水主機性能驗證方法，碩士論文，國立台北科技大學。

[15] 朱正宇，應用灰色系統理論於冰水主機耗電量預測，碩士論文，國立台北科技大學。

[4] K.T. Chau, K.C. Wu, C.C. Chan and W.X. Shen, "A new battery capacity indicator for nickel–metal hydride battery powered electric vehicles using adaptive neuro-fuzzy inference system," Energy Conversion and Management 44: 2059–2071, 2003.

[5] Zaheeruddin and Garima, "A neuro-fuzzy approach for prediction of human work efficiency in noisy environment," Applied Soft Computing 6:283–294, 2006.

[6] Evgueniy Entchev and Libing Yang, "Application of adaptive neuro-fuzzy inference system techniques and artificial neural networks to predict solid oxide fuel cell performance in residential microgeneration installation," Journal of Power Sources 170:122-129, 2007.