基于Dialux结合深度学习技术的办公室智能照明设计  

Design of Office Intelligent Lighting Basedon Dialux and Deep Learning Technology

作　　者：陈穆 CHEN Mu(Chuzhou City Vocational College,Chuzhou 239000,Anhui,China)

机构地区：[1]滁州城市职业学院,安徽滁州239000

出　　处：《兰州文理学院学报（自然科学版）》2024年第6期66-71,共6页Journal of Lanzhou University of Arts and Science(Natural Sciences)

基　　金：滁州城市职业学院教学研究重点项目(2022zdjyxm11);安徽省职业与成人教育学会教育教学研究一般项目(AZCJ2023192);安徽省质量工程教学研究一般项目(2023jyxm1616);安徽省高校哲学社会科学研究重点项目(2023AH052832)。

摘　　要：针对传统的照明设计方案不能有效节能的问题,研究提出一种基于人工神经网络算法的室内光环境优化以及预测方法,既可以减少能耗,又可以提高员工在室内办公的舒适性.结果表明,采用人工神经网络算法进行光控制后,其平均绝对误差、均方误差以及均方根误差依次为0.49445、0.49984和0.48726,经过217次的训练,该方法能够满足预先设定的10~(-2)错误率,从而达到预期的目的,在对办公室室内抽取4个采光点进行预测时,通过模拟得出的最大照度误差为21 lx,低于人类肉眼可感知的50 lx.综上所述,说明该办公室室内光环境优化实用效果很好且稳定性很强.Aiming at the problem that traditional lighting design schemes cannot effectively save energy,a research proposes an indoor lighting environment optimization and prediction method based on artificial neural network algorithm,which can reduce energy consumption and improve the comfort of employees working indoors.The results show that the mean absolute error,mean square error and root mean square error of the artificial neural network algorithm for light control are 0.49445,0.49984 and 0.48726 respectively,and after 217 training sessions,the method is able to meet a pre-determined error rate of 10-2,thus achieving the desired objective.In predicting the four lighting points taken from the office interior,the maximum illuminance error derived from the simulation is 21 lx,which is lower than the 50 lx perceptible to the human eye.In summary,this shows that the office interior light optimization is practical and stable.

关 键 词：BP神经网络 DIALUX LED调光 光通量 

分 类 号：TU113.66[建筑科学—建筑理论]