Data-driven Identification And Analysis Method Research Of Industrial Cluster

Industrial cluster is a special form of industrial organization,in which various subjects form a complex system through various interrelations.Due to the self-organization and competitiveness of industrial clusters,it is difficult to identify the relevant industrial activities among enterprises.The development of self-organization makes the relationship between enterprises loose,which leads to vicious competition among enterprises in the cluster,which seriously restricts the development of industrial clusters.In the global competitive environment,industrial clusters need to be transformed and upgraded,which requires studying the system structure and operation rules of industrial clusters from the micro level.With the popularization of information technology and big data technology,enterprises attract customers and complete related business activities through Internet information,and it is possible to analyze and research industrial clusters by obtaining internet multiple data.In this paper,starting from the data resources of industrial cluster,combining with the method of machine learning,the industrial cluster will be identified,the relationship among the enterprises in the cluster will be analyzed,and the operation law of the cluster system will be explored.This paper takes data as the core to study industrial clusters,and obtains multi-source data of enterprises through Internet technology as the research resources of expanding industrial clusters.Firstly,the framework model of data-driven industrial cluster identification and analysis is proposed,and how to use Internet information combined with text-mining and machine-learning technology to identify and analyze the internal network structure of industrial cluster is described from a systematic point of view.Based on the theoretical research of industrial cluster,the heterogeneous network graph of enterprises and products in industrial cluster is constructed from the dimension of production relationship and spatial relationship,and the data-driven identification method of industrial cluster was proposed by using the convolution neural network of data filtering and integration of acquired industrial cluster.The feasibility of the method was verified by the data sets of Ningbo and Guangzhou.Then,from the perspective of the enterprise association network in the industrial cluster system,the paper maps the enterprise association network to vector space based on the correlation of enterprise product data,and identifies the internal structure of the industrial cluster by unsupervised learning clustering algorithm.Through the analysis method of geospatial,this paper explores the current situation and characteristics of spatial distribution of cluster enterprises and the spatial relevance of cluster members from the perspective of time and space.Finally,the paper constructs the analysis method of the supply network of industrial cluster from the perspective of geographical proximity,and models the supply network of the cluster from the perspective of the spatial association of the main body of the cluster.Based on the GCN model,different parts suppliers and complete machine manufacturers in the supply network are represented in low dimensional vector space.This paper analyzes the distribution pattern and supply relationship of cluster members by using the relationship of cluster enterprises in vector space.At the same time,an example of injection molding machine cluster in Ningbo area was analyzed.The data results show that the supply network analysis method based on GCN model is effective.