The Research On Offset Press Mechanical Failure Learning And Diagnosis Repair System

With the continuous development of civilization in human society, the requirements to the high quality of printing products have become much higher. In order to meet the high quality printing demands, the vast investment of high-tech have made the mechanical structure of printing machines much more complex and the running speed of which much faster, especially the sheet-fed offset press. In this situation, the printing productivity increased and the printing quality also improved, but once a mechanical fault occurred and could not be diagnosis in time, it would inevitably lead to unpredictable losses. Therefore, the establishment of a printing fault diagnosis system for the modern printing press has faced great challenges.The technology of machinery diagnostic changes quickly. The artificial intelligence, expert systems, fuzzy fault diagnosis technology and that of neural network fault have been put into practical application in many areas. But to the practice of printing presses fault diagnosis technology, especially for those native presses, it is still in a backward state, that estimating by a personal experience; it's not only labor-cost, but also time-cost. The native presses can not adapt to integration and automation of printing press with printing industry developing trends.We are sure some advanced technologies such as intelligent fault diagnosis and fuzzy neural network learning techniques will be certainly introduced into the fault diagnosis system of printing press in the future. There are some theoretical researches in this area by some researchers, but an integrated presses fault diagnosis system has not yet established. In this paper, we did a following preliminary exploration for the establishment of the printing press fault diagnosis system,1,Making the relative comprehensive investigation and a collection of expert experience, and doing a summary and classification of common causes and solving methods to the mechanical failures of the current native offset printing press.2,Classifying the common printing press failures by ways of Access database and establishing a printing press fault database to the common faults and causes by the way of data tables.3,Building the common printing press faults learning system based on work above with Visual Basic program to facilitate the printing press faults learning and training for primary printing operators and in-school students.4,Surveying a dozens of experts and operators in the malfunction questionnaire, and summarizing the probability of common causes of the printing press faults, thereby establishing its abstract and quantitative fuzzy rule base, and developing a primary fault fuzzy diagnosis platform of the printing press.5,Exploring primarily the use of BP neural network learning method to analyze the press fault signals and applying it to the fault diagnosis system on the basis of neural network knowledge.