Simulating environmental impacts on building envelopes using machine learning and advanced data analytics

使用机器学习和高级数据分析模拟环境对建筑围护结构的影响

• 批准号: 536519-2018
• 负责人: Rouhani, Hossein
• 金额: $ 1.82万
• 依托单位: University of Alberta
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2018
• 资助国家: 加拿大
• 起止时间: 2018-01-01 至 2019-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=663752
• 关键词: Simulating; environmental; impacts; building; envelopes

Engineers rely heavily on the use of wind tunnel measurements of model buildings to assess cladding and structural loads, local dispersion around buildings, and pedestrian level winds. Wind tunnel testing is an industry-standard approach that has been widely used since the early 1970's. However, wind tunnel testing can be both costly and time consuming. This project will investigate the feasibility of mining and leveraging data from historical, physical scale wind tunnel measurements to develop and train machine learning algorithms capable of generating synthetic, realistic results without the need for bespoke wind tunnel testing. This capability will allow qualified wind engineers to use 'synthetic' (i.e., machine learning derived) wind tunnel type results to provide early design consultation both quickly and cost-effectively.

工程师在很大程度上依赖于使用风洞测量模型建筑物，以评估覆层和结构载荷，建筑物周围的局部分散和行人水平风。风洞测试是一种工业标准方法，自20世纪70年代初以来一直被广泛使用。然而，风洞测试可能既昂贵又耗时。该项目将研究挖掘和利用历史物理规模风洞测量数据的可行性，以开发和训练机器学习算法，该算法能够生成合成的，逼真的结果，而无需定制的风洞测试。这种能力将允许合格的风力工程师使用“合成”（即，机器学习衍生）风洞类型的结果，以提供快速和具有成本效益的早期设计咨询。