Development And Application Of Product CMF Innovation Design System Based On Deep Learning

In recent years,with China's overall manufacturing industry upgrading and the increasing of consumers' quality demand,the products quality has become a key factor when consumers considering and purchasing commodities.The research on the optimization design of product CMF(Color Material Finishing)has become more intensive,and it has made a big impact on the product design process.At the same time,the artificial intelligence industry develops rapidly,and the related field of computer-aided intelligent design is also fruitful,resulting in a large number of new technologies,new theories and new applications.The goal of this research is to assist the product design process by introducing the method of deep learning.In this thesis,we are taking the design process of headphone products as an example,it discovers the combination of deep learning and product design process-by utilizing the method of deep learning on the product design process,we are able to assist the designers with works foundation and to accelerate the design process.The research includes three parts and applied researches.The first part collects commentary data of headphone products by using big data method and extracts strong related product image adjectives,compiles emotional image scoring program of headphone products and headphone CMF annotation program,obtains corresponding data of CMF and product kansei image of headphones,builds CMF kansei image recognition model of headphone products by training BP neural network,with the help of Kansei engineering obtains the consumer's perception of product emotional images under different CMF material combinations and provides support for product research and analysis.The second part constructs innovative design model of product surface texture pattern through shape grammar and neural style migration network,collects traditional national pattern,codes the pattern and generates pattern configuration frame.Introduce frame pattern and traditional national pattern into neural style migration network,extract the small unit features of national pattern through the migration network,map to the frame pattern matrix,and finally convert the matrix back into pattern to realize innovative design of national pattern.This will provide rapid innovative pattern support for surface texture in CMF-P design for product design.The third part collects pictures of headphone products through web crawler,filters and processes the collected pictures data and constructs the data set of headphone product design.Through the training of the data set,an automatic generation system of headphone design for countering network is generated.With the help of the computer-aided method,the repetitive work of designers in product design is reduced.By replacing part of the creative design process with artificial intelligence,it can relieve the pressure of designers and provide more design reference schemes.Applied research: Provide support for product design process through the above research contents.Lastly,Python GUI is used to draw prototype system for CMF innovative design of products containing various auxiliary function modules of product design.The system is used in the headphone innovative design process.The specific implementation process of above system module functions is verified by an example of headphone product design,and the rationality of the prototype system design is verified.This thesis is taking the aspects of CMF optimization design,pattern texture innovation and product design and styling innovation as research points respectively,trying to utilize the in-depth learning method on the product design process,simplifying designer's work through auxiliary product CMF innovation design process,providing theoretical support and innovation inspiration for designers,as well as providing relevant theoretical research support for further product innovation design and intelligent design.