Research On Product Innovation Design Method Based On Big Data

With the improvement of people's living standards,user requirements gradually change from function to emotion.User-centered methods have received attention,while they still have defects with data collection,product visualization,and decision-making.Automation and intelligence are becoming the trend of product design."Made in China 2025" also encourages the use of big data,Internet+ to enhance product innovation,and big data has become a research hotspot.Product life cycle contains feedback,preferences,and vision,how to obtain them from massive data and applied to product design is a key issue.In this paper,we introduced natural language processing,convolutional neural networks,neural style transfer,and multi-criteria decision-making.Combine text big data and image big data with design knowledge to form intelligent methods.This paper focuses on the acquisition,analysis,and application of big data in product design.The main research contents are as follows:(1)Lack of visualization can result in low predictability and long development cycles in product design.To solve these problems,this paper proposes the KENPI method,which can transfer color of a style image in real time to a product's shape automatically.The KENPI combines Kansei engineering,convolutional neural networks,and neural style transfer.To capture user preferences,predict product semantic,and guide style image choose,we combine Kansei engineering and BP neural networks to establish a mapping model between product properties and semantics.The convolutional neural network-based neural style transfer is adopted to reconstruct and merge color feature of the style image,which are then migrated to the target product.The validity and feasibility of the transfer were verified by evaluating the semantics of the new product and target product.(2)Aiming at the problems of strong expert dependence,low real-time and few data in user requirements acquisition and perceptual evaluation,a Kansei engineering driven by online reviews is proposed.This method based on text big data and focuses on product semantic space.First,online reviews and product parameters are obtained from an ecommerce platform through a crawler.Second,the TF-EPA method which integrated Term Frequency with Evaluation-Potency-Activity was proposed to extract Kansei words.The perceptual evaluation was calculated through cluster and adverb-scoring of online reviews.Finally,a BP neural network was used to construct the nonlinear mapping model between product parameters and semantics.By evaluating the generalization ability of the model,the feasibility and effectiveness of the proposed method are demonstrated.(3)Aiming at the problem that multi-criteria decision-making method is not applicable to perceptual evaluation,this paper proposes the KE-GRA-TOPSIS method that combines Kansei engineering,analytic hierarchy process,entropy,game theory,and GRA-TOPSIS.First,an evaluation system is established by TF-EPA and analytic hierarchy process.Second,we define a Kansei decision matrix to describe the satisfaction of user requirements.Third,the analytic hierarchy process is used to obtain subjective weight.Next,the entropy is employed to obtain objective weights by taking the Kansei decision matrix as input.Then the two types of weights are optimized using game theory to obtain the comprehensive weights.Finally,the GRA-TOPSIS takes the comprehensive weights and the Kansei decision matrix as inputs to rank alternatives.The KE-GRA-TOPSIS takes the decision maker's preference as input and outputs the optimal alternative,which realizes precise and individualized decision-making.(4)This paper develops a big data driven prototype system for product innovation design,which mainly includes three core functions: user requirements acquisition based on text big data,product generation based on image big data,and product alternatives decisionmaking.This paper combines big data with design knowledge to form intelligent and systematic methods,which improves the defects in user demand acquisition,product generation and decision-making.The results show that product innovation design methods based on big data can improve innovation,shorten product development cycle,improve product competitiveness and user satisfaction.