Research On Generating Design Of High-rise Residential Facade Based On Deep Learning

Since Alpha Go shines,artificial intelligence has gradually entered people’s field of vision.In April 2020,the Party Central Committee and the State Council will promote artificial intelligence as an important development direction of the "new infrastructure".All walks of life are eager to try,and the architectural design industry is also actively responding.In June2020,the Ministry of Housing and Urban-Rural Development approved Shenzhen’s first plan to conduct artificial intelligence map review.In theory,artificial intelligence is a collection of algorithms,and "deep learning" belongs to one of them.Although the deep learning technology is quite mature,its research in the field of architectural design has just begun,and there is a lot of room for research and improvement.The development of domestic high-rise residential buildings is mature,with a high degree of standardization,and sufficient and high-quality data,which creates a good basic condition for its connection with deep learning technology.In addition,many real estate companies also cooperate with universities and research institutions to develop intelligent design methods,reflecting the industry’s generative design needs.The facade design of high-rise residential buildings has certain stylized and repetitive characteristics,and it also provides a more convenient and reliable testing ground for the initial application of deep learning in the field of architecture.Explore the deep learning-based high-rise residential facade generation design method and its application strategy,provide a new intelligent design method for the high-rise residential facade design,improve the efficiency of design and plan comparison,and make the practical application of intelligent building design A little contribution is the original intention of this research.In order to achieve the goal of this research,the thesis completed the following research work: The first stage,the research on the related technologies and theories of building generative design,deep learning and high-rise residential facade design,established the necessary theoretical basis for this research;In the second stage,with the help of questionnaires and interviews,the current status of high-rise residential facade design and generation design requirements were investigated,which provided a realistic basis for experimental design.In the third stage,the article proposed a high-rise residential facade generation based on deep learning.Design method,and elaborated the method from the four aspects of generative design preparation,database establishment,generative model construction and generative result evaluation.In the fourth stage,based on the results of the preliminary investigation,three types of high-rise residential buildings were studied and formulated.The surface generative design experiment explores the high-rise residential facade generative design method and generative design strategy proposed in this research in an experimental manner;the fifth stage,the research uses a practical case to study the high-rise residential facade generative design method and strategy proposed in this paper Demonstrate and verify,and put forward relevant application suggestions.Finally,the paper summarizes and prospects this research.The paper analyzes the trade-off analysis of the pros and cons of the high-rise residential facade generation design method proposed in this study,and prospects for further research.The final result of this research is to propose a deep learning-based high-rise residential facade generative design method,and sum up three generative design strategies based on this method: human-machine collaborative design process,generative design innovation techniques,and subjective and objective correlation.Combined evaluation method.On the basis of predecessors,this research has also achieved certain innovations.The main innovations are as follows: First,this research integrates "deep learning" and "high-rise residential facade design" across disciplines,and broadens the use of deep learning technology.The scope of research in the field of architectural design;second,this article elaborates on the combination of Pix2 Pix technology and high-rise residential facade design,which has a certain technical reference value;third,based on the current status of domestic high-rise residential facade design,this article The proposed generative design method is practical application-oriented and has certain engineering practical significance.