Regional adaptivity of electrochromic glazing in Japan and operational improvement in energy saving using machine learning

Regional adaptivity of electrochromic glazing in Japan and operational improvement in energy saving using machine learning
复制标题

日本电致变色玻璃的区域适应性以及利用机器学习改进节能操作

DOI:
10.1002/2475-8876.12272
复制
发表时间:
2022
影响因子:
0.9
通讯作者:
Asaoka Yukiyasu
Asaoka Yukiyasu
中科院分区:
--
文献类型:
--
作者:
Kobayashi Takuma;Hiyama Kyosuke;Omodaka Yuichi;Oura Yutaka;Asaoka Yukiyasu

文献摘要

相似文献

电致变色 (EC) 玻璃通过太阳辐射屏蔽减少冷却负荷。然而,过多的太阳辐射屏蔽会增加热负荷。换句话说,EC玻璃的节能效果取决于建筑的能源性能。本研究比较了静态玻璃和 EC 玻璃在各种条件下的热负荷和冷负荷减少效果,以评估 EC 玻璃在日本的区域适用性。此外,为了最大化效果,我们采用基于机器学习(ML)的操作并评估其效率。使用 DesignBuilder 软件根据日本的标准办公模型进行了参数研究。结果表明,与温暖气候下的低辐射玻璃相比,热负荷和冷负荷减少了 17.1%(宫崎,7 区)。然而,在寒冷气候下(带广,2 区),能量增加为 25.4%,并且效果趋势在 4 区附近发生变化。因此,在预计 3 区至 5 区出现热负荷的日子里,我们在工作时间之前纳入太阳能热能。结果表明,预计可以减少 2-3% 的供暖和制冷负荷,并且可以通过机器学习准确设置运行计划。
Electrochromic (EC) glazing reduces the cooling load via solar radiation shielding. However, excessive solar radiation shielding increases the heating load. In other words, the energy‐saving effect of EC glazing is dependent on the energy performance of the building. This study compares the heating and cooling loads reduction effects of static and EC glazing under various conditions to evaluate the regional applicability of EC glazing in Japan. Furthermore, to maximize the effect, we employ a machine learning (ML)‐based operation and evaluate its efficiency. A parametric study is conducted based on a standard office model in Japan using the DesignBuilder software. The result shows that the heating and cooling loads reduces by 17.1% compared with low‐E glazing in warm climates (Miyazaki, Zone 7). However, in cold climates (Obihiro, Zone 2), the energy increase is 25.4% and the trend of the effect changes near Zone 4. Therefore, on days when the heating load is expected to occur in Zones 3–5, we incorporate solar heat before working hours. The results show that reduction in heating and cooling loads of 2–3% can be expected and that the operation schedule can be set accurately via ML.