| Manufacturing industry is a pillar industry in China.With the proposal of Made in China 2025,digitalization,networking and intelligentization are the basic policies for the future development of China's manufacturing industry.However,with the rapid development of manufacturing industry,the complex relationship of equipment parameters,mutual constraints,not easy to analyze the problems of management personnel also appear,and the setting of equipment parameters is one of the important factors to determine the quality of products,so parameter optimization is an important part of the whole manufacturing process optimization.The complex data relationships can be sorted by using the knowledge graph,which makes it easier for managers to clarify the relationships among them.Therefore,how to use the knowledge graph to monitor the parameters in real time and help managers to make parameter optimization decisions is also one of the research hotspots of manufacturing process optimization.The first step to build the knowledge graph of manufacturing process is to process the data.However,the structured data in the manufacturing process may be missing or sparse after being cleaned or due to some factors.How to use the existing sparse data or missing triples to construct a complete and accurate knowledge graph is also one of the main problems encountered in the modeling process of the whole manufacturing process.In this paper,the generative adversarial neural network is introduced to solve the problem of sparse and missing knowledge graph in the construction process.The generative adversarial neural network is used to generate new data,so as to construct accurate knowledge graph.The parameter optimization task can be seen as adding a new node(changing parameter)to the knowledge graph,linking this node to the knowledge graph,and then observing which nodes the node will link with and which parameters it will influence,so as to assist managers to make decisions and optimize parameters.Analysis operator at the same time operating logs,the company's internal experts to investigate the parameters optimization of record,international experts online meetings between data such as knowledge map completion task easier,but these are structured data,so how to make use of unstructured data knowledge map completion task is one of the main problem in this paper.The main work of this paper can be summarized as follows:(1)Taking steelmaking industry as an example,this paper proposes a knowledge graph construction method for manufacturing process parameter optimization based on(2)the background of relevant industries and the application status of knowledge graph,so as to assist managers in parameter optimization(3)In manufacturing process parameters optimization of knowledge map construction encountered in sparse,the problem of the missing data,this paper generated against the neural network and knowledge reasoning solved the problem of data sparseness,loss of data,and in the severstal steel strip test data provided by the company to focus on the experiment,and accuracy than other models will be raised by 3.4%.(4)For unstructured information cannot be directly applied in the manufacturing process of knowledge map completion task problem,this paper,by using neural network and graph of cooperating neural network structured processing unstructured data,and published on severstal company strip data set and UCI plate defect data set,The results showed an improvement of at least 9.7% over some commonly used models. |
PDF Full Text Request