公开(公告)号：US20220318443A1

公开(公告)日：2022-10-06

申请号：US17614291

申请日：2020-10-28

Applicant: SOUTHEAST UNIVERSITY

Inventor： Junyan YANG ,  Daijun CHEN ,  Yi SHI ,  Xiao ZHU ,  Huaxing SHENG ,  Beixiang SHI

IPC: G06F30/13 ,  G06F30/27 ,  G06N3/04

Abstract: The present invention discloses an artificial intelligence (AI)-based method for generating a building block in a commercial district, and relates to the field of AI urban design. According to the present invention, geographic information data of a target district and surrounding districts is acquired to construct a three-dimensional space sand table, then design conditions in various planning files and local legal regulations are translated and extracted to generate a three-dimensional building block of the district, and then a training sample library of three-dimensional contour lines of the district is constructed, a machine learning model is loaded to generate three-dimensional building heights of the district and optimize building forms, to generate a plurality of schemes for the building block of the district, and finally simulated display of the schemes and display of scheme indexes are performed by using a holographic display device, and the schemes are outputted. For the problem that a previous generation of AI technology generates a large quantity of invalid schemes, the present invention realizes the AI-based generation of a plurality of valid schemes, resolving the problem that the previous generation of AI requires a long time and large manpower to screen out a valid scheme, and improving the design efficiency for planners.