| With the integration of big data and industry,the era of intelligent manufacturing,industrial big data technology and data analysis applications have become the key elements to enhance manufacturing productivity,competitiveness,and innovation capabilities.They are driving product intelligence and intelligent production processes.Intelligent management,intelligent service,new business model and new model,supporting the transformation of manufacturing industry and building an important foundation for the open,shared and collaborative intelligent manufacturing industry ecology.Focusing on the research of industrial big data platform technology and data analysis technology,this topic conducts a detailed investigation of industrial big data platform technology,and compares the characteristics of industrial big data and Internet/commercial big data,and proposes a complete set based on Hadoop system.The construction of industrial big data platform,including data collection center,data storage center,data computing center,data application center,and the design,implementation and implementation of the functional architecture,technical architecture and deployment architecture of each center.Based on the industrial big data platform,the relevant production data is imported,and the intractable defects in the TFT-LCD production process are analyzed using big data analysis and machine learning.In this analysis process,XGBoost,clustering and other algorithms are used,and a big data analysis method of hierarchical downward analysis is proposed innovatively.First,the abnormal step is located,then the abnormal line body is retrieved,and finally the line body is analyzed.The root cause of the anomaly is integrated,and a series of equipment parameters of the line body are integrated and analyzed to find out the core influence factors.Finally,the abnormal parameters are analyzed in depth,and the relationship between the parameter interval and the product defect is explored,and the parameter suggestion interval is given.Yield,lower yield,this method can be extended to other industrial applications. |
