A study of fault diagnosis of nuclear power plant clock system based on evidence theory

The clock system of nuclear power plant is used to provide unified accurate time signal to downstream system. In order to keep the system in high reliability, a proper fault diagnosis method is needed to be studied. This paper first discusses the fault diagnosis theory based on D-S Evidence Theory. It gives the fault diagnosis model of clock system based on signal flow model and Bayesian network. Then it uses the Bayesian network to get the fault recognition frame by searching algorithm. A method of acquiring BBA (Basic Belief Assignment) of fault reason is proposed in consideration of system reliability, the knowledge and experience of experts, and self-study of experts. Finally, it proposes the complete method of fault diagnosis and runs an experiment to verify the validity of D-S diagnosis method.