BWR online monitoring system based on noise analysis

A monitoring system for during operation early detection of an anomaly and/or faulty behavior of equipment and systems related to the dynamics of a boiling water reactor (BWR) has been developed. The monitoring system is based on the analysis of the “noise” or fluctuations of a signal from a sensor or measurement device. An efficient prime factor algorithm to compute the fast Fourier transform allows the continuous, real-time comparison of the normalized power spectrum density function of the signal against previously stored reference patterns in a continuously evolving matrix.The monitoring system has been successfully tested offline. Four examples of the application of the monitoring system to the detection and diagnostic of faulty equipment behavior are presented in this work: the detection of two different events of partial blockage at the jet pump inlet nozzle, miss-calibration of a recirculation mass flow sensor, and detection of a faulty data acquisition card. The events occurred at the two BWR Units of the Laguna Verde Nuclear Power Plant.The monitoring system and its possible coupling to the data and processing information system of the Laguna Verde Nuclear Power Plant are described. The signal processing methodology is presented along with the introduction of the application of the evolutionary matrix concept for determining the base signature of reactor equipment or component and the detection of off normal operation conditions.