Research On The Method Of Data-Driven Concept Generation For Product Innovation Design

With the breakthrough and development of information technologies such as mobile internet,big data,and cloud computing,huge amounts of data are accumulating rapidly.How to use these rich data resources to improve the ability of designers to acquire knowledge,expand the wideth of knowledge acquisition,and increase the efficiency of exploring knowledge association,so as to inspire designers to think out of box to design ideation,and enhance the creativity of designers in the process of conceptual design,which is of great significance to product innovation.In the highly open-ended and uncertain conceptual design stage,designers need to expand the knowledge search space and use their imagination to combine their own internal knowledge and external knowledge to solve design problems,realize product innovation to improve the market competitiveness of enterprises.Data-driven design method can help designers to process and mine data,use large-scale data to assist designers in acquiring knowledge related to design problem from a wide range of space,use information technology to establish the relationship between conceptual knowledge,and especially adopt concept network to mine the new,unexpected,weak or unknown relationship between designers' existing knowledge and new knowledge,so as to support designers to perform creative design activities.However,the theory and application of data-driven product innovation design is still in the initial stage of exploration,and the existing research lacks the framework research to support data-driven conceptual design to guide the development of data-driven methods.Moreover,in-depth research is needed on the establishment of associations,semantic distance calculations,and design information capture of concept network,and there is still a lack of research on data-driven methods to support scheme design.Therefore,this dissertation studies the method of data-driven concept generation for product innovation design,which is supported by the state key program of national natural science of China “Research on key problems in the innovative design theory and method of electromechanical systems”.Firstly,this dissertation analyzes conceptual design process,data science and design creativity,and proposed a framework of data-driven design concept generation.Then,in view of the two stages of concept generation and scheme design,the date-driven method is studied to inspire and assist designers to design ideation,and to generate new design concepts;the concept network based method is proposed to provide inspiration and design information for designers to support concept generation,and then a novel method is proposed to push knowledge for designers to support scheme design from the perspective of product structure information.The main research works and contribution of this dissertation are as follows:(1)This dissertation analyzes the application of date-driven method in conceptual design,and proposes a framework of data-driven design concept generation.By analyzing the conceptual design process and the pattern of concept generation,the core of concept generation process is positioned at the two stages of concept generation and scheme design.Then,the application of data-driven method in conceptual design is analyzed by integrating design theory,data science and creativity association theory,which leads to the research of data-driven conceptual design using the method of network science.Finally,based on design creativity,semantic association and design synthesis,the overall framework of data-driven concept generation for product innovation design is proposed.(2)A data-driven concept network construction method is proposed.This method can provide inspirational stimuli for designers and inspire them to produce novel design concepts to solve design problems.Firstly,conceptual knowledge is mined from a large number of text documents,and concept associations are established.On this basis,conceptual knowledge is divided into close,moderate,and remote distance associated concept,and then a novel concept network is constructed.The study further explores the combination between different-distance concepts in order to generate novel concepts expectantly.Finally,the concept network and concept combination are provided to designers as inspirational stimuli,which cover different distances of stimulating information to inspire design ideation.(3)The construction and analysis method of function-structrue concept network is proposed.This method integrates function and structure information to build concept network.Through network analysis and retrieval,it expands designer's knowledge space,provides designers with design information or inspiration,and promotes them to conceive novel design concepts.This research analyzed the function-structure mapping,and extracts the functional information related to product descriptions from the text document based on the characteristics or patterns of function knowledge representation,meanwhile extracts the structural information,and then constructs the function-structure concept network.After analyzing and confirming that the network is a scale-free network,a retrieval method of explicit and implicit associations between network nodes is proposed to push concepts to designers based on the network property and graph embedding algorithm.(4)Structural and semantic information retrieval is proposed to support the stage of scheme design.This method retrieves knowledge containing potential structures and components through semantic analysis from the perspective of structural information,and then pushes it to designers to support scheme design.This paper analyzes the characteristics of the scheme design process,determines to implement support for scheme design from the perspective of structural information.Then constructs the structure vocabulary set and extends it to a more abundant structure feature.Through semantic analysis and clustering calculation of vocabulary in structural features,knowledge representation can be realized from the perspective of structural information.Finally,according to the description text of innovative concepts,knowledge with semantic relevance to concept description is recommended.(5)Based on the data-driven conceptual design method proposed in this paper,the basic architecture and application process of data-driven conceptual design assistant tool prototype system are studied,and the assistant tool prototype is designed,which includes three main modules,namely,word-level concept network module,function-structure concept network module and data-driven scheme design module.