Application of fault detection and identification (FDI) techniques in power regulating systems of nuclear reactors

Application of failure detection and identification (FDI) algorithms have essentially been limited to identification of a global fault in the system, and no further attempts have been made to locate subcomponent faults for root cause analysis. This paper presents Kalman filter-based methods for FDI in power regulating systems of nuclear reactors. The attempt here is to explain how the behavior of the states, residues, and covariances can be interpreted to identify subcomponent faults. An alternative to the Kalman filter-the risk-sensitive filter-is also introduced. Comparison of its performance with the Kalman filter-based FDI algorithms is studied. All simulation studies have been carried out on postulated faults in the power regulating system of heavy water moderated, low pressure vertical tank-type research reactors.