Research On Pleasant Emotion Evaluation And Design Method For Product Form

Improving product attractiveness and usability has been a new research direction for product design nowadays.The functional aesthetics pursued in product design is gradually replaced by emotional experience,and the new usability is transformed into positive emotional demand.Pleasant emotion,as a type of emotion deserves considerate study among positive emotions,has gradually become the topic of new usability in emotional design.This study established a quantitative evaluation model for product pleasant design and conducted a detailed pleasant design system,using emotional theory and cognitive psychology.Firstly,to effectively solve the problems of subjectivity,small amount of data,and poor timeliness in extracting consumers’ Kansei image and needs in traditional Kansei Engineering,the deep learning technology to mine complex data and extract pleasant Kansei image was used.Secondly,to address the problems of small sample size and difficult measurement of consumer decisions,this paper established a mapping relationship between users’ visual cognitive characteristics and product form complexity,and combined aesthetic pleasant and PAD emotion measurement to form a composite pleasant evaluation model.Then a neural network regression model applicable to small sample data was developed,which complemented the shortcoming that a small sample size of perceptual data that cannot meet the requirements of machine learning.Finally,this paper used the harmony search neural network to select the pleasant form harmony library,and established a correlation analysis model by combining Quantification Type I and integer linear programming,and guided and optimized product design decision according to the results.For pleasant emotional design of product,the theory was thus expanded to inspire new research ideas.The contents and innovations of the dissertation are shown below.(1)In response to the dilemma that relying on general Chinese dictionaries for sentiment analysis in Kansei Engineering research often leads to low resolution and low credibility of results,a fusion approach to construct a product domain sentiment dictionary was proposed.From the perspective of user needs,the BERT-LDA model was introduced into the positive emotion corpus of online reviews to mine the perceptual representation of pleasant from the deep semantic structure.The sense similarity was considered to calculate the emotional degree of pleasant words,and 51 pleasant Kansei vocabulary sets were extracted from the disordered product corpus data.After refining the meaning of the pleasant Kansei vocabulary set,three important Kansei image factors were obtained,thus,the mining and analysis of specific Kansei images in product design under big data was completed.(2)The cognitive complexity that affects the pleasure of product form was studied,and the hesitation fuzzy linguistic term set algorithm was used to defuzzify the user’s perceptual evaluation.We also combined the complexity information axiom theory in design,innovated the cognitive complexity calculation model,and measured the perceived complexity of users’visual form.To overcome the current situation that traditional weight calculation methods are not beneficial to design-decision,the complexity weighted entropy calculation method was improved,so that the results of complexity principle weight were more conducive for observation and evaluation.We proposed a quantitative method for users to perceive the pleasure of product form,and according to the complexity and pleasant of users’ cognition,a neural network prediction model for small data size was established.It solved the problems of difficulty in collecting Kansei data and inability to meet the data size of a neural network training model,and realized a small sample-size prediction model.Compared with the traditional prediction model,the accuracy of the proposed model in this study can be improved by more than 40%,and up to 62%.(3)To reduce the subjective interference in Kansei Engineering,an evolutionary algorithm based on harmony search and the neural network was established,thus the most pleasing product samples(harmony library)were intelligently selected.After the product form was split,a Quantification Type 1 integer programming optimization model was innovatively proposed,and the associated mapping between design elements and pleasant emotional images was established.Finally,the optimal pleasant combination was output to guide design decisions,improve the reliability of product design.This study used cutting-edge machine learning and sentiment analysis techniques combined with optimization algorithms for effective research on pleasant design evaluation methods.It achieved the systematic,intelligent and innovative decision creation of product form pleasant design,which was applicable to the mining of customer needs and optimization of product pleasant design in various product categories in big data context.Results showed that our methodology can maximize innovation development capability of companies and satisfy pleasant emotional needs of customers.It provided theoretical support and ideas creation for research on specific sentiments within emotional design.