Research On Product Innovation Design With Generative Adversarial Network

New social,economic,cultural and technological trends have brought new challenges to designers.Consumers now tend to want products that show their individuality,meet their emotional needs,and even participate in the design themselves.And with the upgrade of industrial production technology,enterprises also have the ability and motivation to produce small-lot,personalized goods.At the same time,industrial upgrading and making products with more added value are also the needs of our industrial and commercial development.The use of data intelligence analysis technology is a very promising method to improve the efficiency and quality of results of designers,and among them,generative adversarial network is a kind of neural network that is very easy to use by enterprises and individuals among the deep learning technology under the intelligent analysis technology.In this thesis,firstly,the basic theory of intelligent analysis and neural network is studied in depth,and further research work is carried out based on it.The main work of the whole thesis is as follows:(1)To construct a picture of the combination of data intelligence analysis and the design field through the deconstruction of the design process and historical trends,so as to investigate the impact of generative adversarial networks on the innovative design of products and the positioning of deep learning techniques in the design process,which will be useful for predicting future trends.(2)Experimentally explore the advantages of generative adversarial network-aided design and go through the complete design process to verify the involvement of generative adversarial network in the design.The case study investigates the design of a decorative table lamp,using an open dataset as well as crawling data,training a neural network model based on a Python platform,and aiding the design through the generated results of the neural network model.(3)A summary of the advantages and disadvantages of generative adversarial networkassisted product innovation as reflected by the research content.Through this study,it is found that the application of product innovation design based on generative adversarial networks can effectively improve design efficiency and design quality.The method can continuously generate design samples for designers’ reference or inspiration,and is also expected to enable consumers to directly select product intent in the future,ultimately achieving a more efficient design process.However,there are some limitations of the method,such as the influence of dataset selection and the training effect of the model,which need further research and exploration.This study provides support for applying generative adversarial networks to the design process,and provides a theoretical framework with theoretical as well as practical implications for combining the design domain with data intelligence analysis methods.