The Design Problem Solving Based On Data-driven

This research is granted by National Program on Key Basic Research Project(such as 973 Program)and supported by design practice project.According to the design problem's uncertainty,With intellectual products as supporter of theory research,put forward the way of design problem solving by data-driven,The model,strategy and method of design problem solving by data-driven are constructed,which is of methodological significance to the theoretical research and design practice.The subject of this research is the design problem solving by data-driven which starting from the design problem's uncertainty,to establish the certainty approach of problem solving,and has the characteristics of multi disciplines."Data" is a concept in the field of information science."Data-driven" is a term in the field of computer science,"Problem solving" is a concept of interdisciplinary,in this paper,problem solving based on the design theory:Based on the basic rules of design activities,to establish strategies and methods of design problem solving in the point of view of design pattern,design scene and product link.The research ideas of this paper:1)Research the design problem solving in the theoretical framework:On the basis of information theory,Introduce data to eliminate uncertainty,So as to build a rational solution process,Based on the information theory,bounded rationality theory,art creation theory to conduct for design problem solving by data-driven,Refined a design method of limited perceptual derived of bounded rationality theory.2)Based on design phenomena,refine and solve problems:Based on the rule of design process,the key design scene is extracted,and the way to solve the design problem is realized,Fitting model of artificial neural network based on the cognitive link phenomenon of products,to provide procedural cognitive design problems.3)Using experiments and case studies to demonstrate the related phenomena and problems:Through the experiment to explain or demonstrate related problems or phenomena,and then obtain the correlation method,the design method of practical design case to demonstrate operability,improve the application value research.In this article,the research contents and results are as below:First,the design pattern of the design problem solving by data-driven.From the point of view of the times,this paper researched the evolution of the product iteration,and hold that the "data" is an important feature of the intelligent era.The data is divided into "concrete data" and "abstract data",which are corresponding to the two patterns of product design,namely "product-data-product" pattern and "data-product-data" pattern,The two patterns were formed their own innovation cycle,through the key nodes of"product type" and“product elements" constitute the innovation model of product design by data-driven and use practice case to demonstrate design process under the mode of two design patterns.It is found that the product and the data form a complex network,Design of complex network driven design problem,the essence of data-driven design problem solving is the discovery of new links based on the product elements and new nodes based on product type.Second,through the data semantics,driving the design scene to solve the design problem.Based on the design process,the key design scene is constructed,the design problem is defined and the design problem is solved.In this paper,we select three key design scenarios,namely,the prototype design scenario,creative design scenario and brand design scenario,In order to drive the design scene,based on the semantics of data,the function of the design scene is realized through the semantics as a mediation.In the prototype design scene,based on functional features and styling features,we integrate them to obtain the product prototype.In the scene of creative design,metaphor is used to obtain the formal features and the meaning features,achieve the creation of ideas based on the similarity of design ideas.In the scene of brand design,based on the concept of the product to achieve the mapping of the brand characteristics to the product characteristics,through the mapping model to achieve a consistent description of the semantic features and styling features,the completion of the brand characteristics of the product,complete product of the brand characteristics.The feasibility of the design scene function is demonstrated by the design experiment and the design case,semantics is the medium of design scene by data-driven,and design scene is a way to solve the design problem of data-driven.Third,the fitting model of the design problem solving of data-driven based on the product link.There is a link in the product cognition,the data link between the product and the user in the product design pattern,the user interacts with the product through the data.There is a semantic link phenomenon in product design scene,Product and stimulus compose design chain,product and association compose cognitive chain,product and product can form a complex link network through semantics,semantic and semantic can form a complex link network also through products.Based on the product link phenomenon,the artificial neural network theory is introduced to build the design model of design problem by data-driven.It is considered that the design model of design pattern by data-driven is adaptive resonance theory neural network model,which can realize the self organization cycle of product and data with the function of self adaptation,the product can update iteratively based on the data.and it is considered that the design scene model is BP artificial neural network based on the characteristics of design scene operation,The model can adjust the input weights based on the expected value,and can obtain a satisfactory solution.This two kinds of neural network models are integrated by using "product type" and "product element" as nodes,and the adaptive neural network model with 8 layers is constructed,in the design model,the data is used as input and the product is used as output,to solve the problem of product design is a process of competitive learning network,based on the forward feedback to achieve the desired value of each level,the output of each layer based on feedback to make correction,correction is a dynamic adjustment of the rational process of design iteration.It is obtained a good cognitive structure for the design problem solving by data driven.Fourth,the strategy and methodology of the design problem solving by data-driven.In the design and creation of artistic sensibility,the design principle of limited sensibility is summarized,it is considered that the design scheme is a satisfactory solution,and the satisfactory solution has the stage,which is the result of the stage under the specific conditions,the specific time series,the specific user cognition.Based on the relationship between the product and the data in the design model,the design framework of "product to data" and "data to product" is constructed,carried out"product to data" representation,and it is concluded that the innovation of "data to product" is mainly embodied in the innovation of product type and element,the innovation of product service and experience,and use a case to validate the design framework by the development of a intelligent system.The data is the standard of each node in the design process,the product type is a "filter" of the data to the semantics,and the product element is a "filter" of the design scene to the product design”,the semantic link system is constructed by means of semantics,and the feature is used to implement the product complex system,super semantic driven design scene,group semantic driven creative design scene,individual semantic driven brand design scene,the semantic link system obtains the product complex system through the design scene,and bring about product generation.Finally,the reliability of the method and procedure of the design problem by data-driven is demonstrated by the case of the intelligent product system.In this paper,we studied the design problem by data-driven,obtained the pattern,the model,the strategies and the method about design problem solving,and brought uncertain design problem solving activities to relative certain procedures and methods,and maintained the status and application of data in product design.It is of great significance to the design of intelligent products with data characteristics,and it is a useful supplement to the existing design theory,and it is an attempt to innovate the theory of design in intelligence era.