- 學位論文
應用基因-灰色系統理論於冰水主機耗電量預測
Application of Genetic-grey System Theory to Predict Chiller Power Consumption
摘要
本研究採用灰色理論結合基因演算法來建立冰水主機耗電模型，利用基因演算法搜尋出整體最佳解之優點，依照灰預測原始數列的特性，快速搜尋出相對應之最佳α值，使建立之GM(1,N)預測模型達到最佳準確度。結果說明，最佳基因-灰色預測之方法可解決最佳α值的選擇問題，所建立的最佳GM(1,N)預測模型準確度比傳統方法(α設為0.5)高。 中央空調系統之耗電量與冰水主機運轉效率的高低有著密切關係，不同於線性迴歸與類神經預測需大量數據建模，利用灰色預測只需少許數據（最少四筆）就可以進行的優點下，可快速建立主機GM(1,N)模型以便進行預測。將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，在短期預測上準確率高達99％以上，透過此方法找出耗電量最低，性能最佳之冰水主機並做冰水主機排序最佳化(Optimal Chiller Sequencing簡稱 OCS)，使冰水主機工作於最大效率以達到節約能源之目的。
並列摘要
This study employed the grey theory incorporating genetic algorithms to set up energy consumption models of chiller units, search the corresponding optimal α value, based on the characteristic of the gray predicting original series and the merits of searching global optimum solution by the genetic algorithm, and reach the best accuracy of the GM (1,N) prediction model. Results showed that the method of the optimal gray prediction with the genetic algorithm could solve the problem of the selection of the optimal α value and the GM (1,N) prediction model had better accuracy than the traditional method (α set at 0.5). The energy consumption of the central HVAC systems is closely related to the operating efficiency of the chiller units. The grey prediction required only fewer data (at least four data) to rapidly set up the GM (1,N) model of the chiller units to predict the energy consumption, while the linear regression and the neural prediction required a lot of data to set up the model. Considering four data of the inlet and outlet temperatures of the chilled water and the cooling water at the operating chiller units, the grey prediction had quite accuracy. Compared with the actual power consumption, the accuracy of the grey prediction reached over 99% at a short-term forecast. By the said method, we could find the least power consumption and the highest performance of chiller units, arrange the optimal chiller sequencing (OCS), and operate the chiller units at the highest efficiency for energy-saving.