以快速傅立葉轉換建立建築耗能預測模型

Building Energy Prediction Model using Fast Fourier Transform

周思維 ； 蔡尤溪

《冷凍空調與能源科技雜誌》 92期 (2015/04) Pp. 50-59

摘要

本論文以快速傅立葉轉換作為研究工具，分析建築耗能各項參數，將各項參數由時域轉換為頻域，成功分析出各項參數具備的週期性以及頻譜圖，找出各種頻率的振幅以及相位值，藉此建立週期性參數的模型，並且歸納出建築尖峰耗能的運轉週期。透過傅立葉轉換，處於時域的資料能夠轉換為頻域的資料，各項參數皆可以表示為多組的弦波疊加而成，而各個弦波都有屬於該弦波的頻率、振幅以及相位值，如此便能夠分析出週期性參數的相互關聯，並且建立該參數的預測模型。本論文以決定係數（R^2）值判斷模型的準確率，並且採用實際建築作為案例，利用eQUEST建築耗能模擬軟體建立3D立體模型，以輸出全年度的逐時資料，作為分析參數準確率的判斷依據，結果顯示無論是天氣資料、建築熱負荷或是建築耗能…等參數，R^2值皆高達九成九以上。

關鍵字

傅立葉轉換 ； 快速傅立葉轉換 ； 建築能源 ； 建築耗能預測

並列摘要

Fast Fourier Transform was applied to analyze the parameters of building energy. Energy data were transformed from time domain into frequency domain. The periodicity and the frequency spectra of the parameters were successful analyzed. Amplitude and the phase difference of the parameters in frequency domain were determined. The periodic parametric models of building energy were then built, and the periodicity of building peak energy demand was discussed. The Fast Fourier Transform not only can transform time domain data into frequency domain, the time variation of the parameters can be simulated by recombination of sinusoidal functions. Moreover each sinusoidal function has its own frequency, amplitude and phase. The correlations between the periodic parameters can be analyzed to derive the prediction model of each parameter. Annual hourly building energy data of a case building were obtained using a 3D building energy computation model eQUEST. Fast Fourier Transform of energy data were tested against the computation. The Coefficient of determination (R^2) was found to be higher than 99% for parameters including weather conditions, building thermal load, and energy consumption.

並列關鍵字

Fourier Transform ； Fast Fourier Transform ； Building Energy ； Building Energy Prediction