Real-time Monitoring Technology And Intelligent Data Analysis Of Robot Production System

With the "Made in China 2025" strategy put forward,the transformation of traditional industries to intelligent manufacturing has become a trend,and industrial intelligence will be the key to the future intelligent manufacturing system.Industrial big data technology is one of the key technologies of the intelligent manufacturing standard architecture,and it is also a key element to promote intelligent manufacturing and improve manufacturing production efficiency and competitiveness.Because the traditional data analysis method can not solve the complexity of manufacturing environment and process well,this paper introduces industrial big data technology to analyze and research the robot production system and its industrial big data.The main research includes the following aspects:(1)The remote monitoring system of robot production is constructed.According to the characteristics of robotization,small batch,multi variety and multi type in actual production,multi-source heterogeneous database is designed.Multi thread technology and hybrid network structure is adopted to realize the visualization and real-time monitoring of robot production system.(2)A data mining algorithm library system for industrial big data is developed.According to the characteristics of industrial big data,such as multi-modal,high-throughput and strong correlation,the open and modular structure is adopted to design the general interface standard,which realizes the high cohesion and low coupling requirements of the system, and improves the flexibility and expansibility of the system.(3)The non-parameterized clustering model,unbalanced data flow mining model and data mining algorithm are mainly studied.According to the characteristics of complex and difficult to obtain parameters in intelligent manufacturing scene,a non-parameterized clustering model is proposed,which can be used for automatic clustering analysis of industrial big data.In view of the unbalanced phenomenon and concept drift in industrial big data flow,this paper puts forward the unbalanced data flow mining model,which can effectively identify various types of concept drift in data flow while solving the unbalanced problem in data flow.In this paper,the non-parametrized clustering model and the unbalanced data flow mining model are tested in the simulation and actual scenarios respectively.The results show that compared with the common clustering algorithm,the accuracy of non-parametrized clustering model in simulation experiment and real experiment is improved by more than 20% and 30% respectively;compared with the common data flow mining algorithm,the accuracy of unbalanced data flow mining model in simulation experiment is improved by more than 7%.