Research On System Of Data Acquisition And Quality Monitoring Based On SPC

In the “Made in China 2025” national plan which unveiled by China's State Concil in 2015,“Quality first” is one of the five basic principles,that the quality of products has become a national focus of objects,and the primary objective of the production process in manufacturing enterprise.Therefore,how to effectively monitor the quality of the production process to ensure the quality of products is an important research problem in manufacturing enterprises.Statistical Process Control is the main technology for manufacturing enterprises to monitor the quality of products.It can monitor whether the production process is abnormal according to the image features of SPC control chart,and the reason of the abnormal production process can be judged by the recognition of the SPC control chart pattern.In recent years,researchers use a variety of intelligent algorithms to identify the SPC control chart pattern recognition,due to fewer samples of the current production process control chart pattern,and the support vector machine can effectively solve the problem of pattern recognition under the condition of small sample,so the support vector machine has become the main research tool of control chart pattern recognition problem.The main research contents of this paper are as follows:1)Acquired SPC control chart pattern data by simulation experiment,and extract the features by their statistical features and shape features,studied and analysed of the principle of support vector machine and its multi classification method,selected the Directed Acyclic Graph based on support vector machine as the mulit classifier,and used Particle Swarm Optimization to optimize the parameters of support vector machine,then carried out the corresponding simulation experiment to comparative analysed;2)In order to further improved the classification efficiency and accuracy of the control chart pattern,put forward two kinds of improvement methods: combining the original data with the feature data by feature fusion,then through the principal component analysis algorithm to further decreased the dimensionality of fusion features,so that to extracted the data features which greater impacted on classification,and finally improves the classification accuracy and recognition efficiency;improved the particle swarm optimization algorithm by enhancing the active search ability of particles,to solved the fault of easy to fall into “the local optimal value”,which improved the accuracy of the classifier recognition.Extract the data feature of great influence on the classification,improve the classification the accuracy and efficiency of recognition;particle swarm algorithm was improved by enhancing its ability to active search,to solve the “easy to fall into local optimum defects”,to improve the accuracy of classifier recognition.Through the simulation experiment proved that,the recognition accuracy and recognition efficiency has corresponding improvement based on the improvement of the identification and optimization algorithm,the recognition accuracy can reach more than 95%,it can meet the basic production requirements and effectively prevent the occurrence of abnormal production.