Joint study on anomaly detection and remaining useful life prediction of ex-core neutron detector in nuclear power plant under interference factors

The remaining useful life (RUL) prediction of the ex-core neutron detector is needed to provide helpful references for reactor operation, enable equipment condition monitoring, and enhance condition-based maintenance (CBM) in nuclear power plant (NPP). Most equipment RUL prediction systems fail to integrate anomaly detection methods to determine the onset time of anomalies and neglect the interference factors generating the outliers. We demonstrate the effectiveness of combining the virtual redundancy anomaly detection model with the RUL prediction model in overcoming these issues, based on operational data from a set of real-world intermediate range detectors (IRDs). We also propose an outlier detection method to remove the outliers generated by interference factors. Moreover, a RUL prediction model and a deterioration trend tracking prediction model have been developed.