Research On Product Form Evolution Design Driven By Online User Evaluation

User-centered product design is the crucial design concept that product manufacturers insist on.Therefore,narrowing the gap between product design and user needs is always an important goal of design research.Kansei Engineering is a technology that can transform users'needs into design elements through scientific methods.There are three problems in the research process of Kansei Engineering:First of all,in the process of constructing the Kansei image space of products,the extraction of Kansei image vocabulary often comes from magazines,manuals or reports,so the real emotional needs of consumers cannot be fully reflected,and also the number of sampling is scarce.Secondly,in the process of refining the Kansei image dimension,surveys combined with factor analysis and clustering algorithm is usually used to realize dimensionality reduction,which leads to slow data update and low collection efficiency.Finally,how to deduce the Kansei image of products as design elements,build the quantitative mapping relationship between product design elements and users' Kansei images,and improve the accuracy of training and prediction are the key issues in the research of Kansei Engineering.The mapping relationship model established by linear and nonlinear models used in traditional Kansei Engineering usually has a large error.In addition,it also lacks the analysis of the overall shape coupling of products composed of many design elements.In view of this,this research combines the theory and research framework of Kansei Engineering,uses the text mining and natural language processing to analyze the online evaluation texts of users on the e-commerce platform,and obtains the users' real Kansei needs information from it,which could construct product Kansei image space in a faster and more efficient way.This research establishes the relationship between design elements and Kansei needs based on support vector regression,and builds the evaluation and prediction model of product form image.Finally,taking the interactive genetic algorithm as the basis,the product form evolution design system is constructed,and the product image form evolution design is realized,so that the product form design plan that meet the user's preference is obtained,which broadens the theoretical boundary for the research of complex product form design,and also reduces the cost and time of product development.The main research work is as follows:(1)The evaluation text of online users is obtained.Writing crawler functions with Python software to obtain online user review information.Based on the two major e-commerce platforms of Taobao and Jingdong,a total of 48 electric bicycle products,were selected for review crawling.Among all selected objectives,each has more than 15 reviews,and the product form meet the new national standard.A total of 288,510 words of review texts are crawled,and stop words that are not strongly related to the semantic meaning of the texts are removed according to the text preprocessing rules.The Jieba tool is used to perform word segmentation processing on the review texts.(2)The user image requirements are accurately analyzed based on text mining.The TF-IDF keyword extraction method of Natural Language Processing was used to analyze the user's perceptual image keywords in the online user reviews,and the collection of 260 product Kansei vocabulary was completed.Through the integration of focus group and KJ method,the total vocabulary was merged and simplified into 67 perceptual image vocabularies.In order to identify its semantic characteristics,use the WordNet tool to retrieve semantic synonyms and antonyms from the results to form a semantic network.The representative Kansei semantic vocabulary in the network is analyzed by applying sociometric status method to construct the user's Kansei image semantic space.To analyze the relationship between emotional image need and emotional satisfaction,the Fuzzy Kano Model is used to identify and classify user needs,then the emotional image needs that are expressed as attractive are filtered out,which are"simple","technological","exquisite","dexterous" and "fashion" Kansei image needs.(3)A product image evaluation model is established on the basis of user emotional preference.The evaluation gird method is used to investigate the morphological characteristics of products that are attractive to users,and 542 product morphological elements are obtained,and similar elements are merged through the KJ method to construct a small product morphological data set,and 28 product morphological design elements are obtained.According to the Support Vector Regression,the mapping relationship between product form elements and user's Kansei image is established,and the product multi-dimensional Kansei image evaluation model that conforms to the user's emotional preference is constructed.Through comparing the error results of traditional BP neural network,the advantages of the established evaluation model are effectively verified,and the accuracy and performance of the model are also effectively verified,thereby assisting the designer in accurate market positioning.(4)A product evolution design method is established on the basis of user's Kansei image.Grounded on the constructed electric bicycle product form database,the design elements are binary coded.According to the analyzed Kansei appeals of users,the Fuzzy Analytic Hierarchy Process is used to calculate the weight of Kansei appeals.Taking user's Kansei preferences as the evaluation target,the product image morphology evolution design system is established based on the interactive genetic algorithm.In order to avoid the individual fitness deviation caused by the ambiguity and uncertainty in user's evaluation process,and to reduce user fatigue,the user evaluation hesitation time is adopted to improve individual fitness.The Kansei image evaluation model based upon Support Vector Regression for automatic evaluation could further reduce and alleviate the fatigue generated in the user evaluation process.The Rhino software is used for three-dimensional modeling of product morphological design result,and the fuzzy comprehensive evaluation method is used to test the user satisfaction for the product design scheme.Hence,the evaluation results put the effectiveness of the product image morphological evolution design method constructed by this research to the proof.Based on the construction of the semantic space of user's Kansei image and the evolutionary design of product image form,this paper studies the product form that meets the user's preference.The results obtained are valid to guide enterprises in the innovative product form design and assist designers in product design.This paper is of great practical application value in enterprise product research and development.Furthermore,it is also an effective supplement to the research on the theory and method of Kansei Engineering.