Kalman filter-based maximum a posteriori probability detection of boiling water reactor stability

A diagnostic system has been developed to determine the global system stability characteristics of an operating boiling water reactor (BWR) using the average power range monitor (APRM) signal as an input. A Kalman filter-based monitor is used to identify the stability characteristics of a BWR using an "M-ary" hypothesis testing scheme. Each hypothesis, i.e., system model with slightly different stability characteristics, is represented by one of several Kalman filters operating in parallel. In addition to calculating the current estimate of the system\´s states, the monitor also produces the a posteriori probability that each filter calculates an accurate state vector estimate given the measurements. The best estimate of the system stability corresponds to the Kalman filter producing the maximum a posteriori probability (MAP). Results suggest that a significant time savings, when compared to stability diagnoses performed via a more established technique (autoregressive modeling) requiring 30 min to 1 h worth of data, can be obtained using the MAP BWR stability monitor. MAP monitor diagnoses are typically performed in 20 to 30 s. Thus, the MAP monitor represents a novel approach for the detection of BWR instabilities.