Research On Cooperative Neural Network Ensembles For Control Charts Pattern Recognition

With the progress of production technology,consumer demand is increasing.This demand not only means that the increase in demand,more demand for quality improvement.Quality management is an important method for improving the competitive advantage of modern industrial production.In the process of modern industrial production,the stable process is an important factor affecting the quality of the product.Statistical control of the process of quality control chart,often used to monitor the stability of product quality,and the traditional control chart is no longer meet the needs of modern large-scale production.With the help of advanced computer information processing technology,the application of artificial intelligence technology in industrial process control to achieve the industrial process of quality control of real-time,accuracy is the current domestic and foreign experts and scholars to study one of the direction.This paper summarizes the domestic and international research status and development trend of control chart pattern recognition in the field of quality management in the process of modern industrial production,and introduces the basic concepts of statistical process control and the basic principles of quality control chart.The basic theory of neural network and its generalization and integration theory,co-evolution and other related theories are expounded and analyzed,which provides theoretical support for the research of this paper.By analyzing the shortcomings and shortcomings in the current pattern control method of quality control chart and combining with the characteristics of artificial neural network in dealing with complex classification problems,this paper proposes a method of neural network integration design and training by using co-evolution theory.Through the analysis of the generalizationerror of the neural network,the neural network learning algorithm and the co-evolutionary algorithm are combined,and the error of the integration network of the individual network is used to realize the difference of the individual network.The structure of the individual neural network is automatically determined in the learning process,the accuracy of the individual network is maintained,the structure of the neural network is automatically determined by the construction method,and the stability and generalization ability of the integrated learning system are improved.Finally,the Monte Carlo quality feature data simulation method is used to generate the quality characteristic sequence similar to the actual production process.MATLAB2012 a is used to program and control the six basic pattern recognition networks.The simulation results show that the trained CNNE model has strong And its performance is obviously superior to BP neural network and RBF neural network traditional integration method,also better than Bagging and Adaboost method.