Research On Quality Prediction Module Of MES In Intelligent Factory

Intelligent factory is an important part of the development of intelligent manufacturing,and manufacturing execution system(MES)is an important hub to realize the information integration of intelligent factory.Therefore,in order to promote China's intelligent manufacturing 2025,the research of MES in intelligent factory is very important.In Intelligent Factory MES,the dimension and depth of data mining and analysis are developing towards full data.Conventional technology of MES can not deal with the pre-warning and global monitoring of massive real-time data.At the same time,as the lifeline of manufacturing industry,product quality also has higher requirements for the real-time accuracy of the whole sample quality prediction.Based on this,the quality prediction of MES in intelligent factory is studied in this paper.(1)Analysis of MES and its quality prediction system in intelligent factory.In the context of intelligent manufacturing,traditional MES faces the problems of technological system reform and emerging technology integration.This paper analyses the technical framework of MES,designs the functional structure and system integration of MES in intelligent factory,and establishes the functional structure of its quality prediction system.(2)Establishment of quality prediction model driven by large data.Aiming at the characteristics and real-time accuracy requirements of multi-feature and few samples for quality prediction in intelligent factories,a quality prediction framework adapting to large industrial data is established firstly.Then,considering the high efficiency of BP neural network in processing massive data and the accuracy and rapidity of XGBoost algorithm in processing large dimensional data with multi-feature and few samples,product quality based on BPXGBoost hybrid model is designed.Quantity predictive control model.(3)Simulation and verification of MES quality prediction model in intelligent factory.To verify the effectiveness of the proposed quality prediction method.Using the product quality data of multi-feature and few samples,the BP neural network,XGBoost model and BP-XGBoost hybrid model are simulated.The comparison results show that the BP-XGBoost hybrid model has more robust learning effect than the single model,and has a good effect on real-time prediction accuracy.(4)Development of prototype system for MES quality prediction in intelligent factory.Based on the functional structure design of MES system and its quality prediction system in intelligent factory,the basic modules and functions of MES system are analyzed,and the design and implementation of database and prototype system are carried out.