Application of autoregressive modeling techniques in BWR noise analysis

Abstract Experience on development and application of an autoregressive (AR) modeling technique in BWR noise analysis is described. Based on the improvement of the AR model using an associate matrix, which specifies signal transfer paths in measured signals, BWR noise in reactor core and control systems is analyzed. In the reactor core, void and recirculation flow noise sources, which are most dominant in BWR plants, are quantitatively evaluated by the improved AR model. Also, ordinary and partial coherence functions between radially located neutron detector signals are estimated to clarify spatially distributed noise sources. In the control system, the open-loop and closed-loop responses are identified from steady-state noise data. Furthermore, a failure detection and isolation method for the control system is developed on the basis of the improved AR model.