在照明節能分析中，能夠簡便又精確的預估照明設計耗電是非常重要。北美照明 學會 (IES) 及日本照明節能分析中，照明用電密度 (UPD) 是必備的基本要項。然而該值均 採單一固定值，雖然簡便但缺乏因應各種照明條件，而能較精確的評估。因此本研究利用理 論推導找出耗電影響因素，經統計迴歸分析，得到最具影響的四個變數；照度、光源效率、 室指數、器具效率， 利用實驗計畫法模擬 324 解析模型，分別建立直接照明及半直接照明 之預測模式，其複修正判定係數分別為 0.92、0.80。 本模式並與理論式比較，其複相關係 數達 0.98，證實本模式是可信賴的。 本模式僅包含四變數因此使用簡便，且可隨變數之不 同預測照明設計之耗電量，因此較單一固定值為合理。

　　　　　Among the design and evaluation criteria for energy conservation, the unit power density(UPD) is a key element to calculate the power budget of a room. The disadvantage of the base UPD is that there are too many various tasks to be entirely listed. Besides even the same type of task there is possible different base UPD due to its different illuminance levels and illuminating designs. This research is trying to establish a simple UPD calculation model which is able to deal with the actual lighting power consumption more simply and accurately. To establish this model, a theoretical formula is used to find out the UPD influential factors, then a computer software is used to simulate 324 samples with direct lighting and semi-direct lighting designs. A statistical computer software is also used to establish a linear regression model which will cover four variables. The study results have 0.98 statistical correlation with the ones by theoretical formula. Its shows that this model used to predict the power consumption is reliable and more accurate than the model suggested by the IES.