Fault Identification and Forecasting in NPPs Using Computer-based Operator Aid Tool

This paper introduces a computer-based operator aid tool (COAT) for fault identification and forecasting in nuclear power plants (NPPs). Fault is identified by combining auto-regressive integrated moving average (ARIMA) model and error back propagation (EBP) algorithm. The patterns of unknown faults are then fed to an identifier based on the semi-supervised learning (SSL). Transductive support vector machine (TSVM) is used to cluster the type of unknown fault. To forecast future states of NPPs, a hybrid network combining ARIMA and ANN is developed. Faults in Bushehr nuclear power plant (BNPP) are examined. The Results are in good agreement with the FSAR.