由短期監測數據預估長期空調負載變化趨勢及節能改善方向

Using short Term Monitored Data for Long Term Prediction of Air-Conditioning Load and Energy Saving

張滌淞(Ti-Sung Chang)

指導教授 ： 李魁鵬

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)

https://doi.org/10.6841/NTUT.2008.00325

摘要

本論文藉由探討氣象資料與建築耗能的關連性，試建構一可預測長時段建築耗能的數學模型，以期達到用電量預估與了解節能改善成效的目的。
首先透過類神經網路的模擬，以短期監測數據延伸預測長時段的建築耗能，提出用電契約容量最佳化的建議，以降低電費支出；同時也透過eQUEST動態耗能分析軟體分析該建築物的耗能情況，了解該建築實際耗能改善的空間。最後透過兩者的比較分析，提出節能改善方向與成效的說明。
在論文內提出影響建築耗能氣象三要素(乾球溫度、含濕量及輻射強度)的預測手法及各種氣象資料(如TRY、TMY、TMY2、WYEC、WYEC2、日本標準年、中國標準年)製作的方法，同時也說明了本論文研究使用方法的詳細步驟與相關運用軟體程式的原始碼，希望透過連續性的研究，提供後續研究者一些想法與方向，也希望後續研究者能跳過摸索期，能運用到前人的經驗，而將更多寶貴的時間與精力，運用在更多的實務與創新上，做出更有幫助的貢獻。

關鍵字

類神經網路 ； 灰色理論 ； 動態耗能分析軟體(eQUEST) ； 能源預測 ； 能源模擬

並列摘要

In this paper, by studying the relationship between the weather data and building energy consumption data, a mathematic model is established to predict the long term building energy consumption and power demand, and understand how efficient it is when improving the energy saving.
First, by applying the simulation of the neural network, the monitored short-term data tells the tendency of long-term building energy consumption. This helps to decide the best cost-saving power demand when having a contract with Tai-Power Company so to reduce the electricity bill. Moreover, with the eQUEST analysis, the energy saving potential can be figured out. At last, with the comparison of both analysis, the effective energy saving measures will be clearly identified.
The forecast of three factors of weather (dry bulb temperature, humidity ratio and density of radiation) and the processing of weather information (e.g. TRY、TMY、TMY2、WYEC、WYEC2、Japanese standard year、Chinese standard year) are introduced in this paper. Besides, the detailed steps of the model and the programming source code are attached for further reference.

並列關鍵字

ANN(Artificial Neural Network) ； Grey System Theory ； eQUEST ； Energy Prediction ； Energy Simulation

參考文獻

[10] Kawashima M, Charles E Dorgan, etc. Optimizing system control with load prediction by neural networks for an ice-storage system [J], ASHRAE Trans, 1996,102(1): 1169-1178.

[13] Bishop C M , Neural Networks for Pattern Recognition[M], London: Clarenolon Press, 1995.

[14] Mackay D C , Bayesian Nonlinear Modelling with Neural Networks[R], Modelling phase Transformations in steels, University of Cambridge Programme for Industry, 1995.

[23] Mackay D C , Bayesian Interpolation[J], Neural Computation, 1992.

[24] A. Mechaqrane / M. Zouak, A comparison of linear and neural network ARX models applied to a prediction of the indoor temperature of a building, London, Neural Comput & Applic (2004) 13: 32–37.