Machine learning-based vertical wind profile estimation for high-rise buildings by using building morphological parameters

Vertical meteorological conditions are the main external factor affecting the energy simulation of high-rise buildings. In urban areas, just one universal vertical wind profile is provided for all buildings' energy assessment and simulation. However, since the various morphological features of buildings and environmental conditions, vertical wind profiles from different wind directions can be various even in the same city. To get more accurate building energy prediction and simulation, it is necessary to develop precise and fast vertical wind profile estimation methods. Therefore, this study developed machine learning-based methods to generate vertical wind profiles with building morphological parameters as inputs. To validate the proposed method, this study conducted a case study with the wind tunnel data in Hong Kong. The results suggest that machine learning-based methods can effectively reflect the variation of vertical wind profiles from different wind directions in urban areas, and support vector regression shows the best performance. © (2023) by the European Group for Intelligent Computing in Engineering (eg-ice). All rights reserved.