Research On Key Technology Of Diagnosis And Maintenance System For Complex Equipment

At present, the requirement of complex equipment maintenance can't be met because of some shortage, such as the deficiency of diagnosis and maintenance knowledge in early maintenance, singleness of diagnosis method and its limitation in detecting incipient failures. Contraposing the above problems, the problem of acquirement of diagnosis and maintenance knowledge of complex equipment is considered from the view of product lifecycle. Then, related application research is carried out and corresponding system is developed. The main contents are as following:Firstly, the diagnosis and maintenance philosophy of complex equipment which guides the research on corresponding system is expatiated. And then the system's conceptual model, architecture and logical structure are established.Secondly, in order to change the actuality of closeness of knowledge between product design and maintenance, diagnosis and maintenance-oriented knowledge model is established based on PDM. In the model, diagnosis BOM is used to organize design knowledge needed in fault diagnosis, which provides a new method for the acquirement of design knowledge needed in fault diagnosis.Thirdly, diagnosis knowledge organization model of complex equipment is established. Product structure tree is used to organize diagnosis knowledge in the model. The method can acquire knowledge from diagnosis BOM directly, and lays the foundation for automatical construction of diagnosis model.Fourthly, Bayesian networks are structured automatically by utilizing deep knowledge of diagnosis knowledge organization model. Deep knowledge derives from product design domain, so structured Bayesian networks can provide early maintenance for equipment. Then, the integrated strategy combining rule-based reasoning with Bayesian networks reasoning is given.Fifthly, fault evolution process of complex equipment is researched. The importance of identifying incipient failures in recessive abnormal work status is indicated. Then the meaning , characteristic , work background and flow of fault pre-diagnosis are expatiated. Based on support vector machine, the prediction model of characteristic parameter is established.And fault pre-diagnosis method based on support vector machine and rough sets is presented.Finally, automatic door of train is taken as the application object to research the development and implement of diagnosis and maintenance system. The whole scheme and service flow of diagnosis and maintenance system are introduced, along with the methods of implementing the system functions. Lastly the system's testing result is showed.