Research On Key Issues And Application Of Remote Cooperative Fault Diagnosis

Nowadays, the complex equipments are propelled to large-scale, full automation,high running speed, which leads the fault detection and diagnosis (FDD) into adilemma under traditional mode for the high cost, low efficiency and etc. Therefore,remote, cooperative, intelligent methods are introduced into FDD domain and becomethe research focus recently. Try to address the issues arisen in the FDD for thecomplex system under e-maintenance environment, this dissertation take the flexiblemanufacturing system (FMS) as the application object, proposes an alternativescheme on the remote cooperative FDD, which includes framework construction,supporting techniques, system realization and application study.First, based on the analyzing of research background of remote FDD, theliterature review and the enterprises survey, this dissertation presents problemdescription and put forward a remote cooperative FDD framework for the complexsystem under e-maintenance environment, which is titled with E-CFD (e-CooperativeFDD). Try to address the acquirement, transmission, analysis and decision of multi-sources diagnostic information, a center-distribution FDD framework is constructedbased on the process integration of device to e-business (D2E). By analyzing thetheoretic principle and key technologies, gives the logical structure and functionalmodel of E-CFD, which provides a way to the effective integration of FDD systemand e-maintenance environment.Second, for the task decomposition and control within remote cooperative FDD,this dissertation gives a hybrid algorithm based on relation model and Bayes network.The relation modeling method is introduced to give solution of FDD pathprogramming in coarse level. Based on incidence matrix and Bayes network, thegraph theoretic method is given to address the task decomposition in detail. To realizethe diagnostic resources configuration under multi-constrain conditions, the optimalscheduler model is discussed. Third, the game-theoretical information fusion model is introduced to addressthe problem of the conflicting multi-source fault diagnosis decisions in the process ofcooperative diagnosis. The basic element, functional model, process model and fusionalgorithm are discussed in detail. For the specific problem arisen in the FDD domain,this dissertation take the multi-source decision information as the players in thecooperative FDD, and give a fusion algorithm based on entropy to process thedecision problems in quantificational ways and give the optimal diagnostic decision.Fourth, try to increase the intelligence of the remote cooperative FDD, anANN-based reconfigurable cooperative behavior simulation model is discussed.Based on the analyzing of the behavior intelligence in the product life cycle, thisdissertation gives a tri-layers framework constituted of computing component,cooperative rule, and reconfiguration setting. Within such a model, this dissertationgives the construction methods of intelligent FDD system.Last a prototype system titled UN-E-CFD is under development to give asupport platform for the research issues. The application to FMS and robot proves thepreviously mentioned models and algorithms are effective in practice.