Identification of pressure control system dynamics in BWR plant by multivariate autoregressive modeling technique

Dynamics of a BWR plant pressure control system were studied using stationary noise data for development of a plant diagnosis technique. A multivariate autoregressive (AR) modeling technique has been widely used for similar purposes. However, this cannot be applied to the pressure control system, since the system contains highly coherent signals and its partial transfer mechanism possesses a fast time constant. The present study solved these difficulties by introducing an associate matrix, which described a priori relationships between objective signals, into the conventional AR model. Using this modified AR model, internal (open loop) and closed loop transfer functions were identified. Their validity was confirmed by comparing with actual transient test data. These results show the effectiveness of noise data for evaluating not only partial component dynamics in the pressure control system but also whole system dynamics, such as a plant stability. (author)