基於物聯網通訊協定的兩階段燈光自動調節機制

Abstract

燈光對人有很大的影響，適當的燈光有穩定身心及提升工作產能等好處，然而一方面燈光卻也是建築用電中最主要的消耗來源之一，因此如何調控燈光，在這兩者取得權衡，成為了重要的研究目標。過往對於燈光的調光程度與照度間的假設，並未能準確表達兩者之間的關係，且須事先於量測點量測最大調光程度時之照度，若欲量測地點或燈光位置有所變動，則須重新量測，有不夠彈性的缺點，故本論文提出了兩階段燈光自動調節機制，在第一階段使用了類神經網路，透過訓練後的類神經網路，以距離和調光程度作為輸入因子，只要知道調光程度和燈光至量測點之距離，便可以預測照度。接著第二階段再根據第一階段的預測結果，透過模擬退火法找出各燈泡的最適調光組合，本論文提出的燈光自動調節演算法在照度預測上的平均絕對誤差僅有以往方法的12%，且因為在照度的預測上更為準確，能找到更符合需求的燈光調光組合。

Lighting impacts human’s activities. Appropriate lighting has lots of benefits, such as stabilizing human’s mood and enhancing productivity. However, lighting accounts a large proportion of power consumption in buildings. So research on how to strike a balance between user satisfaction and energy consumption is getting important. In the past, assumption of the relationship between light’s dimming level and illuminance is not absolutely right, and previous method has to measure illuminance of user position in light’s full brightness. It makes lighting system inflexible. If there is any change in light’s or user’s position , the method does not work. To improve the drawbacks aforementioned, we propose a two-phase autonomously adaptive lighting mechanism. In the first phase, we’ll use neural network. Once the neural network is well-trained, it can predict the illuminance with distance and dimming level as inputs. In second phase, we’ll use Simulated Annealing to get the best dimming set of lights based on trained neural network. Compared to previous methods, the mechanism proposed in this paper get only 12% error and because of more accurate prediction in illuminance. Our method can get more appropriate dimming set of lights.