A Study of Artificial Neural Network on the Optimization of Interior Lighting Design

• Main Authors: Chien-fu Tsai, 蔡見福
• Other Authors: Horng-ching Hsiao
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/7t9463

碩士 === 國立臺灣科技大學 === 電機工程系 === 94 === The purpose of this thesis is to study the optimization of interior lighting design by using artificial neural network. Typical classrooms are chosen to be the lighting zones for this research. The influential input variables of classroom lighting quality include: luminaire type, luminaire disposition, luminaire orientation and luminaire spacing. This study proposes a method of using neural network to predict three indicators of lighting quality: average illuminance, unified glare rating, and lighting uniformity; and to evaluate overall lighting quality design with consideration of power consumption indicator. The results show that by applying back-propagation neural networks to learn and generalize the internal mapping rule of classroom lighting design input variables and indicators of lighting quality, it is feasible to predict lighting indicators and to evaluate lighting quality.