Timing Data Processing And Quality Prediction System For Injection Molding Process

In recent years,injection molding industry has developed rapidly in China due to broad development prospects and huge market demand,occupying an important position in Chinese light industry and becoming one of the pillar industries of the national economy.However,compared with developed countries,Chinese injection molding industry still has a big gap,there are low added value,low technical content and other problems.With the continuous development of big data,Io T,artificial intelligence and other information technologies,China has put forward the concept of smart factory and digital workshop,and applied the Internet of Things technology to the injection molding industry.Currently,injection molding informatization mainly collects a large amount of industrial data through new sensors.Due to the characteristics of fast generation frequency,large amount of data and time sequence in the process of injection molding,such systems are difficult to efficiently process,store and apply injection molding data.In the intelligent aspect of injection molding,how to use artificial intelligence technology to improve the efficiency of injection molding production is becoming more and more important.For the complicated injection molding process,all kinds of parameter adjustment and disturbance may cause the quality defect of injection molding products,so the rapid and effective quality defect prediction of products is helpful to improve the quality inspection efficiency of products,reduce the economic loss caused by production failure.Based on the above problems,this thesis combined with the production characteristics of the injection molding process and the Internet of things,big data,such as technology,designed and implemented for injection molding process of temporal data processing and quality prediction system,effectively promoted the injection molding industry data sharing and exchange.It is importance to reduce the injection molding product defective rate,promote the intelligent injection molding industry information.The research contents of this thesis are as follows:（1）Studying the basic concept and working principle of MQTT communication protocol,designing the transmission module of injection timing data based on MQTT protocol,and realize low overhead and low broadband occupancy of message transmission in injection molding scene.Meanwhile,Infux DB timing database is used to store the timing data transmitted by MQTT protocol efficiently.（2）Aiming at the problem of huge working data generated in injection molding process and the time consuming of processing a large amount of data,a visual architecture of sequential data was built by using Echarts to realize the visual reading,modification and online monitoring of massive injection molding data.（3）Combined with the specific characteristics of the injection molding process,pretreatment and feature extraction of historical time series data,put forward XGBoost product quality prediction algorithm based on SMOTE overssampling to achieve the quality prediction of injection molding products.（4）Designing and implementing a platform system for injection molding timing sequence data processing and quality prediction based on distributed architecture.On the basis of Spring Boot framework,nginx,Dubbo,Zookeeper and other technologies are combined to decouple each module to improve the scalability and stability of the system.