Research On Intelligent Monitoring Technology For Printing Process Quality Based On SPC And Neural Network

With the development of informatization, printing products as a medium to carry information, its quality is more and more important, and the printing enterprises pay more attention to the quality control for printing matter. In the printing process, the fluctuation of product quality is inevitable with the influence of various factors, to realize the high quality and high efficiency, the printing process must be controlled in stable state, so the process control of production quality is particularly important. On the other hand, the digital printing workflow improves the level of informatization and automation greatly, and the state of printing process cannot just rely on human factors, otherwise it will block the improvement of automation and quality management for printing enterprise seriously.This paper applies SPC method and the BP neural network to quality control of printing process, and intelligent monitoring technology of printing process quality is researched.Firstly, according to the characteristic of printing process, SPC method is applied to quality control of printing process. Through the analysis of the characteristic of printing quality, the objects of SPC method are determined. The application scheme of control chart is proposed, x ? R control chart and x ? Rs control chart are selected, and control chart of printing quality is divided into the whole area and single ink area, then show the application of histogram and control chart through example.Secondly, pattern recognition for control chart is researched based on the BP neural network. First, control chart for printing process quality is divided into four basic patterns, as the normal, step, trend, and period, their mathematical model is established. Then, the overall scheme of pattern recognition is proposed, and intelligent recognition method of basic patterns and characteristic parameters estimation of abnormal patterns are researched emphatically. Through the experiments, scaled conjugated gradient training algorithm is adopted, four BP neural network models are established, BP network-1 is used to recognise four basic patterns, another three networks are used to estimate the characteristic parameters of abnormal patterns, in these three models: BP network-2 is used to estimate amplitude of the step, BP network-3 is used to estimate amplitude and length of the period, BP network-4 is used to estimate gradient of the trend.Then, according to the theory of SPC, each kind of abnormal patterns is analysed based on six aspects, as man, machine, material, method, environment and others, then, summarize the typical abnormal causes and diagnostic suggestions.Finally, printing quality management database and diagnosis knowledgebase are established based on the SQL Server2000. With Visual C++ 6.0 and Matlab7.0 as development platform, the intelligent monitoring system for printing process quality is programmed, its functions include management of printing quality data, drawing of histogram and control char, intelligent recognition of process and characteristic parameters estimation of abnormal patterns, and the diagnosis for abnormal patterns. Result of the test is all right.