Research On A Configurable Data Acquisition System For Multiple Working Conditions

This paper takes a company’s aerosol filling production line as research object.To improve the filling production line’s degree of informatization and intelligence,a configurable data acquisition system is designed and implemented.The system is aimed to be adaptable to multiple working conditions.Monitoring of production process and fault diagnosis of production line are realized by collecting data,storing data and analyzing data.Enterprises can improve the product qualification rate by adjusting production in time.The main contents of this thesis are as follows:(1)Combining the demands of company with the actual situation of filling production line,the overall scheme design of data acquisition system is completed.The overall scheme includes data acquisition part,server part and client part,which are respectively used for data acquisition,storage and analysis.The frameworks of data acquisition part and server part are designed.Data acquisition part includes production data acquisition terminal,fault data acquisition terminal and product data acquisition terminal.By selecting different acquisition terminals,data acquisition system can adapt to multiple working lines.(2)Design three kinds of data acquisition terminals.Production data acquisition terminal obtains data through sensors and uploads data through HTTP protocol,then stores the production data to server part.The terminal’s hardware part includes the minimum system,communication module,collection module and storage module.The terminal’s software part includes FreeRTOS real-time operating system,time slice rotation scheduling algorithm and HTTP interface.Fault data acquisition terminal uploads information of different maintenance status through HTTP protocol,then stores the fault data to server part.Product data acquisition terminal uploads the data obtained from PLC through MQTT protocol,then stores the product data to server part.(3)Study the outlier elimination algorithm and design the database on server part.By using distribution graph algorithm,the collected data is preprocessed and the outliers are eliminated,which ensures the accuracy of data.MySQL is chosen as the database in server,and three data tables are designed to store three kinds of data acquisition terminals’ data respectively.The problem of slow query speed caused by massive data is solved by partitioning data tables,and the storage of historical data is completed by transferring historical data to guarantee the performance of tables.(4)The statistical process control method and machine learning algorithm are studied to monitor production process and realize production line’s fault diagnosis.By using univariate control chart and multivariate control chart,the production process is monitored,and the stability of the production process is judged.Then the abnormalities in production process can be found in time.BP neural network and BP neural network optimized by genetic algorithm establish the fault diagnosis model by learning the training set data.The verification is completed through the test set data,then the fault diagnosis of filling production line is accomplished.The contributions of this thesis are as follows: a configurable data acquisition system which can adapt to multiple working conditions is developed,and the collection of production line’s important data is realized.Also,production process is monitored through control charts,and fault diagnosis of production line is realized through BP neural network optimized by genetic algorithm.The research enhances the modernization and intellectualization of the filling production line.