Distributed Dynamic Health Assessment For Complex Systems

In the recent years, the complexity of the maintenance task for large systemshas been increasing because of its role in keeping and improving system availabilityand safety, which leads to the fact that it is more difficult for most current existingcentralized health assessment system to meet the demand. Thus, to solve theproblem, the thesis presents a distributed architecture based on remote methodinvocation and multi-agent techniques for dynamic health assessment of complexsystems which is implemented to provide services enabling the evaluation of thehealth status of complex systems to determine if future missions of complexsystems can be carried out.The whole distributed system is composed of5subsystems, containing theglobal knowledge base subsystem, the distributed monitor agent subsystem, thedistributed diagnosis agent subsystem, the distributed prognosis agent subsystem,the graphical user interface subsystem. The diagnostic function is used fordynamically analyzing the symptons received from the monitor layer to identify thefaulty component of complex systems in certain failure mode by using thecause-symptom rules and the knowledge base, and the prognostic function is incharge of dynamically figuring out the RUL (remaining useful lifetime) of thecomponents by applying the generic prognostic algorithm and the knowledge base,which makes maintenance operators evaluate the future health of the systems toenable the isolation or repair or replacement of components that needs maintenanceactions.In practice, the system based on the Archistic project is running well in thelaboratory. Generic multi-agents technique and the distributed middleware RMImakes the distributed agents run on more different computing platforms byimproving the system availability and Expansibility. In addition, the distributedagents can be deployed closer to the components of the complex system and analyzea great amount of information more evenly and accurately to reduce the number ofuseless removals of devices, which can lower the risk and maintenance cost.