Research On Building Skin Generation Design For Building Energy Efficiency

In the context of sustainability,society has increasingly higher requirements for building energy efficiency.As the main interface between the inside of the building and the outside,the building skin is not only a representative of the overall appearance of the building,but also a key part that determines the interior comfort and energy-saving performance of the building.This paper mainly studies the design method of building skin generation which is conducive to energy saving,and reduces the energy consumption of building indoor air conditioning system and lighting system by optimizing the energy saving performance of building skin.The research takes the effect of the change of building skin form on its energysaving performance as the starting point,and deeply explores the building skin generation design method for building energy saving;systematically summarizes the algorithm and logic of building skin generation,and summarizes the building skin generation algorithm that can be reused Set;quantify the energy-saving performance of the building skin through the simulation of the thermal and light environment of the building,link optimization algorithms to drive the building simulation simulation,and iterative calculations to obtain a set of plans with good comprehensive performance of the building skin.The machine learning method is introduced to construct a supervised learning model based on scikit-Learn to apply to the generative design of building skin,which reduces the time for building skin performance optimization and judgment,expands the creative space of architects,and improves the early stage of building skin,and improves the efficiency of the scheme design.The research focuses on three themes of building skin shape,skin performance optimization,and machine learning applied to building skin generative design.There are six chapters in total.The introduction of the first chapter is the research background,analysis of research problems and determination of research ideas.The transition to the basic theoretical overview of the three contents of related building skin,simulation and optimization,and machine learning in Chapter 2 provides theoretical support for the research.Chapter 3 systematically sorts out the methods and logic of building skin formation,using software tools such as Rhino,Grasshopper,and Ghpython to summarize the building skin generation methods into basic formations,mosaic formations,fractal formations,interference formations,and weaving.There are eight categories of biomorphism,folding biomorphism,force biomorphism and bionic biomorphism,and related cases are listed to illustrate.Chapter 4 takes the mosaic shaping method as an example,refers to the building mass and skin plan of Beijing Greenland Center,sets up the building mass model,constructs the building skin program,uses multi-objective optimization algorithm to drive the simulation to optimize the skin energy saving performance;describes the model in detail The establishment,simulation,optimization algorithm application,data processing,and the parameter setting and experience skills in the screening of the result plan form a complete work flow.Chapter 5 analyzes the limitations of the workflow of shaping,simulation,optimization and screening by repeatedly constructing experimental models of building skin generative design based on multi-objective optimization.Combining the characteristics of computational design methods using algorithms and data generation,machine learning methods in the field of artificial intelligence are introduced.Set up the experimental model with equal gradients,use multi-objective optimization to generate a series of building skin plans,and collect data as a training set.Constructing a supervised learning model based on Scikit-Learn has formed a building skin generative design method combining generative design and machine learning.Finally,through experimental application verification,it is demonstrated that the building skin generation design method for building energy conservation based on machine learning can efficiently assist architects to generate skin solutions with excellent comprehensive performance.Chapter 6 summarizes the research innovations and results,and puts forward the research prospects and expectations.