Application of a lattice filter algorithm to BWR noise monitoring and diagnosis

a method for BWR noise monitoring and diagnosis based on an adaptive lattice filter is presented in this paper. It is specifically intended for applying to on-line real time BWR stability monitoring during reactor operation relatively close to unstable regimes, where fast tracking of any stability state change is essential. The authors present a set of algorithms for estimating stability parameters, i.e. decay ratio and peak frequency, for detecting anomaly occurrences based on the pattern recognition methodology, and for calculating signal DC component and mean amplitude. A unique feature of the method is the unified approach for the signal processing, in that these algorithms are all executed based on model identification using the adaptive lattice filter. The present method was applied to BWR noise data collected during stability tests at Forsmark 1 and 2 in Sweden. The result demonstrates that the present method is capable of fast tracking of stability state changes. The paper stresses that the evaluation of the process state by combining the results from the stability parameters estimation, pattern comparison, and checking the behavior of DC component and mean amplitude of process signals is important for obtaining reliable estimate and adequate information on the stability state, especially when the reactor is operated near instability