The Grammatical Generation Study Of Architectural Spatial Form

This study takes computational design synthesis(CDS)method as the research object.It is a kind of automatic generation method from design conceptions to design solutions,which adopts information technology and artificial intelligence to guide design decision-making.CDS is an important research direction of computer-aided architectural design(CAAD),which has a great potential to enhance the creativity,precision and efficiency of design.These advantages are critical for increasingly complex and sophisticated architectural designs.This study selects shape grammar as the formal language for constructing CDS modeling,which defines how the architectural spatial form is represented,as well as how it is generated.The CDS modeling based on shape grammar is essentially a design process modeling,with merits of both parametric modeling and building information modeling.However,the current CDS researches do not make full use of shape grammar,which increases the search burden and reduces the quality and efficiency of problem solving.As such,based on fully exploiting the rationality and diversity of grammatical generation,the proposed CDS approach combines the grammatical reasoning ability with the algorithmic search ability,with emphasizing the full use of the architects' experience and knowledge as well as the algorithms' computational intelligence in creating a variety of design schemes that meet the expected goals.This study initially focuses on the establishment of CDS modeling,which provides architects with two approaches to embed procedural and characteristic design intentions into the structural framework of modeling and the adaptive definition basis of design variables.The former is realized by modular organization of rules and symbolic representation of rule application sequence,while the latter is achieved by establishing the relationships among contextual information,design criteria and design variables within the rules.By using above approaches,two grammars for the generation of building layout and large railway station form are developed as case studies.The comparisons of their application results illustrate that both approaches can enhance the grammatical reasoning ability and improve the rationality of design schemes,which facilitates an accurate solution space.In order to enable the algorithm to effectively guide the design decision process,this study also explores two kinds of CDS calculation methods.The first one adopts a strategy that continuously corrects the deviation design through small-scale local heuristic search in the grammatical application process,to make the final result satisfy as many constraints as possible.The other algorithm is an advanced genetic algorithm which adapts the genotype-phenotype mapping method to the structural and scale dynamics of the grammatical solution space.It is able to achieve design optimization by directly guiding the complete grammatical generation process.The previous two grammar cases as the calculation modelings,are combined with two calculation methods respectively to constructe the CDS prototypes.Their results demonstrate the feasibility and potential of the proposed calculation methods in achieving design goals.This proposed set of approaches covers the implementation of a complete CDS process ?from function to form,and from form to goal?.They provide the architects multiple means to transform their design intentions into the digitial forms,meanwhile improve the applicability of the search algorithms in solving the architectural design problems.They provide theoretical basis and methods for the new cooperative design pattern combining human intelligence and computational intelligence.In addition,these modeling and calculation approaches have been instrumentalized.They are not only used for design generation tasks,but also used as tools for architects to test their design logics and train their abstract thinking skills,which is helpful for promoting the practice of CDS approaches.