Research On Technology Fusion Foresight Based On Link Prediction Based On Graph Neural Networks

As a prominent feature of recent innovation development,technology fusion has a profound impact on the innovation ability of enterprises and the optimization of industrial development.It is of great significance to grasp the future development trend of technology fusion for the first to consolidate the dominant position or the second to catch up and surpass.As an emerging comprehensive manufacturing technology,3D printing has significant features of fusion and development.Studying its potential fusion patterns in the future is conducive to improving the ability to cope with technical problems and accelerating the development of technology and industry.Based on the analysis of co-occurrence network evolution of the core IPC4(4-bit IPC classification code)of 3D printing technology,this paper constructed a link prediction model based on graph neural networks to predict the potential fusion relationships,and analyzed the development prospect of the potential fusion relationships based on the recent research status of 3D printing technology.The main research work and innovative achievements of this paper are as follows:(1)The core IPC4 co-occurrence network of 3D printing technology was constructed based on association rules,and the evolution of the network was analyzed by combining network density,degree centrality,betweeness centrality and bridging centrality.The result showed that the focus of fusion was focused on computer-aided design related fields in the early stage,then turned to molding equipment and manufacturing fields,and the fusion of material fields has become a new important trend in recent years.(2)The link prediction model based on graph neural networks was constructed,and the validity of the model was tested according to the IPC4 cooccurrence network established previously.At the same time,we compared the model of this study with the traditional link prediction models,and the results showed that the link prediction model based on graph neural network has higher accuracy.(3)Carry out prediction for the established period IPC4 co-occurrence networks,calculate the actual incidence of the prediction results in the subsequent periods to further observe the accuracy and reliability of the model prediction results.The predicted results of the last two periods were integrated into a list of potential fusion relationships,which was followed by analyzing the development prospects of potential fusion relationships based on combining the Louvain community detection algorithm and the recent research status of 3D printing technology.The analysis results showed that there was much room for future development in the fields of disinfection materials and devices which belong to medicine or hygiene sectors,in the field of materials such as polymer compounds,alloys and coating compositions,in the direction of control system and data calculation of 3D printing metal powder manufacturing,in the field of organ manufacturing techniques and materials and so forth.(4)This paper put forward constructive suggestions for the government,relevant enterprises and research institutions.In this paper,association rules and link prediction based on graph neural networks were used to mine the potential fusion information in 3D printing patent data,which was beneficial for industry stakeholders to further optimize the business layout,and also provided references for the theoretical research on technology fusion.