| 摘要: | 當今都市人大部分的時間都生活在人造光源底下，以往的照明沒有特別注重在人的專注力、睡眠甚至使用者的情緒上，每每都著重在色彩表現、發光效率等等，尤其疫情時代的來臨，人們不常到室外活動，有更多時間是待在室內，因此更顯得視覺人因照明極其重要。
本研究模擬室內辦公室環境，分析使用者在不同情境照明下的生心理影響和專注程度之影響，並利用基因演算法設計出不同等效黑視素照度(Equivalent Melanopic Lux, EML)或不同晝夜節律刺激值(Circadian stimulus, CS)以及自訂誤差內的色偏差值(Duv)、平均演色性指數(Ra)等條件限制之情境照明。本研究將延續實驗室既有的照明人因評估技術，招募受試者在模擬辦公室進行靜態的紙本測驗，並加以嘗試改進實驗流程、測驗內容。
本實驗固定所有情境的桌面照度為500 lx，並設計出六種不同色溫不同晝夜節律刺激值和不同等效黑視素照度(Equivalent Melanopic Lux, EML)的光源光譜，並使用其情境進行了三個照明實驗。實驗一的研究目的為比較不同晝夜節律刺激值和等效黑視素照度的影響，使用相同色溫5000 K，三個情境分別為情境一CS值0.4搭配EML：500；情境二CS值0.4搭配EML：300；情境三CS值0.3搭配EML：300。實驗二針對實驗一進行部分修改以增加實驗準確性與流暢性，情境的部分改使用相同色溫3000 K，三個情境分別為情境四CS值0.45搭配EML：340；情境五CS值0.3搭配EML：340；情境六CS值0.3搭配EML：170。實驗三則使用實驗二的實驗方式搭配實驗一的情境進行實驗。
Neurosky腦波儀負責量測和記錄受試者在情境下工作、休息的腦波訊號，並匯入MATLAB撰寫程式進行分析，當中包羅了經驗模態分解法(Empirical mode decomposition, EMD)、希爾伯特轉換(Hilbert transform, HT)、機率密度函數(Probability density function, PDF)以及接收者操作特徵曲線(Receiver operating characteristic curve, ROC curve)等運算，最後利用接收者操作特徵曲線的曲線下面積(Area under the curve, AUC)和作答總題數、正確率作為客觀指標；以主觀問卷所得的評分作為主觀指標。主客觀指標分別匯入SPSS統計軟體進行變異數分析，檢視在三種情境之間所使用的指標是否具有顯著差異，之後進一步探究主客觀指標的情境狀況。
研究結果顯示在客觀指標方面，僅有實驗三中的情境一跟情境三的Beta頻段在成對比較具有顯著差異，在其他實驗中α、β、γ的頻帶下情境間沒有顯著差異；在主觀指標方面，實驗一、三中的情境二(色溫為5000 K，CS值為0.4，EML為300)相比於情境一(色溫為5000 K，CS值為0.4，EML為500)在相同色溫、CS值下，較低的EML對受試者的視覺感受會偏疲勞；實驗二的問卷結果顯示在相同色溫下，較低的CS值與EML會影響到人們工作效率等等；紙本測驗的答題情況也可看出在CS值較高的光源下答題，正確率最高的次數較多。
;Nowadays, urbanites live under artificial light sources most of the time. Instead of paying attention to the effects of lighting on human’s concentration, sleep, or even emotions, people used to focus on the color performance and luminous efficiency formerly. However, in the era of epidemics, people spend more time indoors rather than go to outdoor activities. That’s why human centric lighting become increasingly important.
This study takes the office environment as the experimental field to evaluate the effects of different lighting spectra on user’s physical and mental states and attention level. We used genetic algorithms to generate lighting situations with preset correlated color temperatures (CCTs) but different equivalent melanopic lux (EML) or circadian stimulus (CS). Tolerance of delta uv (Duv) and color rendering index (Ra) are within the custom range. The experiment inherited the procedures designed in the team previously. Participants were recruited to perform the static paper-based tests in the office.
The NeuroSky MindWave Mobile 2 was utilized to measure the users’ EEG signals. After that the signals were collected into MATLAB and going through empirical mode decomposition (EMD), Hilbert-Huang transform (HHT), probability density function (PDF), and receiver operating characteristic curve (ROC curve) analysis. And lastly, the area under the ROC curves (AUCs) and the scores of the tests were calculated to become the objective indices. On the flip side, the scores of the questionnaires were the subjective indices.
The first experiment has three lighting scenarios of different EML and CS combinations with a fixed CCT at 5000 K and illuminance at 500 lx. The CS value is 0.4 and the EML value is 500 in the scenario one. The scenario two has CS of 0.4 and EML of 300 and the scenario three has CS of 0.3 and EML of 300. For the purpose of increasing the accuracy and fluency of the experiment, the second experiment uses the modified experimental content from the first experiment. Three lighting scenarios with a fixed CCT at 3000 K and illuminance at 500 lx, with CS of 0.45 and EML of 340 in the scenario four, CS of 0.3 and EML of 340 in the scenario five, and CS of 0.3 and EML of 170 in the scenario six. The third experiment takes the experimental method of the second experiment and the lighting scenarios of the first experiment.
The results from the questionnaires showed that in the first and the third experiment, the scenario two with lower EML value (CCT at 5000 K, CS at 0.4 and EML at 300) made the visual experience more fatigued compared with the scenario one (with the same CCT and CS). The second experiment showed that under the same CCT, lower CS and EML values affected people′s work efficiency. Additionally, the results of paper-based tests showed a higher accuracy rate for the lighting scenarios with a higher CS value in the third experiment. The current experimental design and the method of data analysis shall be applicable to test other lighting scenarios in the future, even though the analysis of brainwaves in several frequency bands shows no significance in the present study. |
