A suitable wind environment is very necessary for future cities. Its function is not only to provide fresh air, but also to help us release pollutants from industrial products and regulate air temperature and humidity.
To achieve this goal, fist step is redefining urban patterns. These new petterns are created following good-ventilation rules which are curve shaped, multi direction and density comparison. The purpose of these rules are releasing extra wind, spreading wind to Where the air is stagnant and making pressure difference to adapt to different functional areas.
Put these new patterns into the style transfer, mchine generate diverse kind of new maps with different textures. The information these results provide could be transferred to different urban elements like streets, residential communities, public spaces through people’s imagination. To obtain a more completed information patterns which include streets, constructions, greens, open areas information, and with a larger scale, I overlapped them and get a new pattern, to transfer to a full xiongan map. In details, new results began to form some clear network, green parks, infrastruction.
How could we know whether this way could really help construct a better wind condition? Through wind analysis we can distinguish these well-ventilation areas with characteristics like natural formation, muti-direction, open to the main summer wind direction and porous structure.
Machine learning provide us a new way to optimize urban space, and this could be one of the possibilities to design the future urbans with better ventilation.