| Surface assembly production is a key part of the electronic product processing process,which will affect the quality of electronic products and limit the speed of the replacement of consumer electronics.Therefore,it is necessary to conduct intelligent manufacturing technology research for surface assembly production enterprises in order to improve the digitalization and intelligence of surface assembly production enterprises.This paper investigates intelligent manufacturing technologies such as multi-source heterogeneous data fusion,dynamic configuration of production materials and process defect diagnosis for surface assembly manufacturing companies.First,it addresses the problem that various departments of surface assembly production enterprises cannot achieve coordination and cooperation through information sharing.A multi-source heterogeneous data fusion data warehouse based on Web Service technology,ETL(Extract Transform Load)technology and data warehouse technology is designed,which will provide data for the subsequent dynamic configuration of productivity and process defect diagnosis.After establishing a connection between the data warehouse and the transactional databases of each information system through the Web interface,ETL operations are performed on the data in the transactional databases using the ETL algorithm written.The processed data is then stored in the data warehouse in a topic-oriented form,and the fusion of multi-source heterogeneous data in each transactional database is realized.Secondly,to address the problem of uncertain production factors causing a lag in the completion of production orders for surface assembly manufacturers.A dynamic allocation algorithm for surface assembly production considering uncertain production factors based on stochastic simulation,neural network,fuzzy theory and heuristic algorithm is designed,which will be used as an intelligent scheduling algorithm for the subsequent smart manufacturing management system.The model was tested using a test case of a designed surface assembly line,and the results show that the model is effective in optimizing the optimal chromosome adaptation value,energy cost and order completion time,and can accurately predict the production task,task volume and each component placement station for each time period of each production line.Thirdly,for the problem of complex manufacturing process in surface assembly production enterprises,which easily leads to product process defects,a knowledge mapping-based process defect intelligence implementation idea is proposed.And a BERT-RBi LSTM-CRF defective entity extraction model that introduces a residual network structure is designed for the defective entity extraction,which is a key technique in the knowledge graph building process.The results of the comparison test between the improved model and the base model showed that the average values of accuracy,precision and F-value of the improved model improved by about 20.26,0.28 and 0.24,respectively,compared with the base model.The Neo4 j graph database is also used to build a defect knowledge graph based on the defect entities extracted by this defect entity extraction model.This defect knowledge map will be used in the subsequent Smart Manufacturing Management System for defect diagnosis.Fourth,a Intelligent manufacturing management system including dynamic configuration of production materials and process defect diagnosis functions is implemented using C# development language with fused data warehouse and defect knowledge map as data base and dynamic configuration algorithm for surface assembly production considering uncertain production factors as back-end algorithm. |
